Set up and Configuration

We load the dataset

import numpy as np
from IPython.display import display, HTML
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix, classification_report
import pandas as pd
pd.options.display.max_columns = None

#df = pd.read_csv('S:/MHS_Pathfinder/bipolar_prediction/GOLD_&_AURUM_15_04_20.csv', low_memory=False)
df = pd.read_csv('clean_dataset.csv', low_memory=False)

NB: Because lo_HDL has missing values, pandas interprets the whole column as float by default.

Characterisation

# What columns do we have?
print('\n'.join(sorted(df.columns.to_list())))

BMI
BMI_date
BP_date
CHD
CHD_date
CKD3
Chronic_pulmonary_disease
Chronic_pulmonary_disease_date
Coagulopathy
Coagulopathy_date
Congestive_heart_failure
Congestive_heart_failure_date
Deficiency_anaemia
Deficiency_anaemia_date
Diabetes_organ_damage
Diabetes_organ_damage_date
Diabetes_uncomplicated
Diabetes_uncomplicated_date
FH_BPD
FH_BPD_date
FH_LD
FH_LD_date
FH_NOS
FH_NOS_date
FH_anxiety
FH_anxiety_date
FH_any
FH_depression
FH_depression_date
FH_psychosis
FH_psychosis_date
FH_substance
FH_substance_date
FH_suicide
FH_suicide_date
Fluid_electrolyte_disorder_date
Fluid_electrolyte_disorders
HDL
HDL_date
HIV_AIDS
HIV_AIDS_date
LDL
LDL_date
Liver_disease
Liver_disease_date
N_dep_b4
N_man_b4
Neurological_disorders
Neurological_disorders_date
OCD
OCD_date
PD
PD_date
Peptic_ulcer
Peptic_ulcer_date
Peripheral_vascular
Peripheral_vascular_date
Pulmonary_circulation
Pulmonary_circulation_date
RA
RA_date
SSRI
SSRI_b4
SSRI_during
T2DM
T2DM_date
TCA
TCA_b4
TCA_during
TSH
TSH_date
Valvular_disease
Valvular_disease_date
Weight_loss
Weight_loss_date
adhd
adhd_date
age_BMI
age_BP
age_CHD
age_Chronic_pulmonary_disease
age_Coagulopathy
age_Congestive_heart_failure
age_Deficiency_anaemia
age_Diabetes_organ_damage
age_Diabetes_uncomplicated
age_FH_BPD
age_FH_LD
age_FH_NOS
age_FH_anxiety
age_FH_depression
age_FH_psychosis
age_FH_substance
age_FH_suicide
age_Fluid_electrolyte_disorder
age_HDL
age_HIV_AIDS
age_LDL
age_Liver_disease
age_Neurological_disorders
age_OCD
age_PD
age_Peptic_ulcer
age_Peripheral_vascular
age_Pulmonary_circulation
age_RA
age_T2DM
age_TSH
age_Valvular_disease
age_Weight_loss
age_adhd
age_alcohol
age_anxiety
age_asthma
age_ca
age_cannabis
age_cardiac_arrythmia
age_conduct
age_death
age_depression
age_dermatitis
age_diagnosis
age_eGFR
age_ethnicity
age_first
age_first_AP
age_first_MS
age_first_SSRI
age_first_TCAs
age_first_diagnosis
age_first_exposure
age_first_li
age_first_olan
age_first_other_ADs
age_first_reg
age_hyperthyroid
age_hypothyroid
age_last_SSRI
age_last_TCAs
age_last_other_ADs
age_mania
age_migraine
age_other_substance_misuse
age_psych_FH
age_psychosis
age_relationship
age_self_harm
age_sleep
age_smoke
age_stress
age_transfer_out
alcohol
alcohol_date
anxiety
anxiety_date
any_AD_b4
any_AD_during
ap_b4
ap_duration
asthma
asthma_date
ca
ca_date
cannabis
cannabis_date
cardiac_arrythmia
cardiac_arrythmia_date
cohort_end
cohort_start
conduct
conduct_date
death_date
depression
depression_date
dermatitis
dermatitis_date
diagnosis_date
diastolic
dob
dominant
eGFR_date
end_reason
ethnicity
ethnicity_date
ex_time
exposure
exposure_end
exposure_start
first_AP_date
first_MS_date
first_SSRI_date
first_TCAs_date
first_date
first_episode
first_li_date
first_olan_date
first_other_ADs_date
first_reg_date
hi_LDL
hi_ca
hypertension
hyperthyroid
hyperthyroid_date
hypothyroid
hypothyroid_combined
hypothyroid_date
incident_script
last_SSRI_date
last_TCAs_date
last_other_ADs_date
li_b4
lo_HDL
lo_ca
mania
mania_date
mania_type
migraine
migraine_date
ms_b4
ms_duration
olan_b4
other_AD_b4
other_AD_during
other_ADs
other_substance_misuse
other_substance_misuse_date
patid
pracid
psych_FH_date
psychosis
psychosis_date
relationship
relationship_date
responder2
response2_1
self_harm
self_harm_date
sex
sleep
sleep_date
smoke_date
smoker
source
stress
stress_date
suitable
symptom_to_diagnosis
symptom_to_exposure
systolic
thyroid_blood
transfer_out_date
weight
year_exposure
yob

# How many unique patients?
print("Total patients:", df.patid.nunique())
print(df['suitable'].value_counts(dropna=False))

Total patients: 38957
1    31518
0     7439
Name: suitable, dtype: int64

Now we only keep patients that are suitable for the inclusion analysis (suitable==1). Those are defined as patients with >2 years of follow-up after exposure_start.

df_old = df.copy()
df = df.loc[df.suitable==1]
print("New total patients:", df.patid.nunique())

New total patients: 31518

# Count values for a few important columns
for item in ['source', 'exposure', 'suitable', 'response2_1', 'responder2', 'symptom_to_exposure', 'exposure_end']:
    print(item)
    print(df[item].value_counts(dropna=False))
    print(df[item].describe())
    print()

source
AURUM    20910
GOLD     10608
Name: source, dtype: int64
count     31518
unique        2
top       AURUM
freq      20910
Name: source, dtype: object

exposure
lithium       19106
olanzapine    12412
Name: exposure, dtype: int64
count       31518
unique          2
top       lithium
freq        19106
Name: exposure, dtype: object

suitable
1    31518
Name: suitable, dtype: int64
count    31518.0
mean         1.0
std          0.0
min          1.0
25%          1.0
50%          1.0
75%          1.0
max          1.0
Name: suitable, dtype: float64

response2_1
0.0    14785
1.0    11848
NaN     4885
Name: response2_1, dtype: int64
count    26633.000000
mean         0.444862
std          0.496960
min          0.000000
25%          0.000000
50%          0.000000
75%          1.000000
max          1.000000
Name: response2_1, dtype: float64

responder2
0.0    19670
1.0    11848
Name: responder2, dtype: int64
count    31518.000000
mean         0.375912
std          0.484365
min          0.000000
25%          0.000000
50%          0.000000
75%          1.000000
max          1.000000
Name: responder2, dtype: float64

symptom_to_exposure
 0.000000     3251
 0.038330       46
 0.002738       45
 0.019165       42
 0.093087       38
              ... 
 4.098563        1
 23.208761       1
 20.591375       1
-15.641341       1
 14.631075       1
Name: symptom_to_exposure, Length: 11420, dtype: int64
count    31518.000000
mean        10.186286
std         12.650202
min        -30.031485
25%          0.813142
50%          6.557153
75%         15.629706
max        148.963730
Name: symptom_to_exposure, dtype: float64

exposure_end
04mar2019    71
01mar2019    65
26feb2019    64
19feb2019    59
18feb2019    58
             ..
30mar2013     1
19jun1989     1
24sep2007     1
24jan2014     1
09dec1999     1
Name: exposure_end, Length: 7796, dtype: int64
count         31518
unique         7796
top       04mar2019
freq             71
Name: exposure_end, dtype: object

Table 1

features13 = ['age_first_exposure', 'age_first_diagnosis', 'symptom_to_exposure', 'psychosis', 'depression', 'mania', 'dominant', 'sex', 'FH_BPD', 'FH_depression', 'FH_psychosis', 'weight', 'self_harm']

features = """Total, N
Age at diagnosis, median (IQR)
Age at medication initiation, median (IQR)
Years between diagnosis and exposure, median (IQR)
Female, n (%)
First presentation mania, n (%)
First presentation depression, n (%)
Depression dominant, n (%)
Psychotic experiences, n (%)
Self-harm history, n (%)
Smoker, n (%)
Family history for bipolar disorder, n (%)
Family history for depression, n (%)
Family history for psychosis, n (%)
Overweight or obese, n (%)"""
featurs_list = features.split('\n')
Table1 = pd.DataFrame(featurs_list, columns=['Features'])


for exp, exp_df in df.groupby('exposure'):
    data_list = []
    data_list.append(str(len(exp_df)))
    for feature in ['age_first_diagnosis', 'age_first_exposure', 'symptom_to_exposure']:
            Q3 = np.quantile(exp_df[feature], 0.75)
            median = np.quantile(exp_df[feature], 0.5)
            median = exp_df[feature].median()
            Q1 = np.quantile(exp_df[feature], 0.25)
            IQR = Q3 - Q1
            data_list.append("{:.2f} ({:.2f})".format(median, IQR))
    for feature in ['sex', 'mania', 'depression', 'dominant', 'psychosis', 'self_harm', 'smoker', 'FH_BPD', 'FH_depression', 'FH_psychosis', 'weight']:
        if feature == 'sex':
            sum = (exp_df[feature] == 'female').sum()
        elif feature == 'dominant':
            sum = (exp_df[feature] == 'depression').sum()
        elif feature == 'smoker':
            sum = exp_df[feature].isin(['ex smoker', 'current smoker']).sum()
        elif feature == 'weight':
            sum = exp_df[feature].isin(['overweight', 'obese']).sum()
        else:
            sum = exp_df[feature].sum()
        data_list.append("{} ({:.2f} %)".format(sum,sum/len(exp_df)*100))
    Table1[exp] = data_list

display(Table1)

	Features	lithium	olanzapine
0	Total, N	19106	12412
1	Age at diagnosis, median (IQR)	40.82 (24.13)	39.08 (22.36)
2	Age at medication initiation, median (IQR)	46.51 (23.20)	42.47 (23.04)
3	Years between diagnosis and exposure, median (...	7.37 (15.60)	5.45 (13.35)
4	Female, n (%)	11526 (60.33 %)	6858 (55.25 %)
5	First presentation mania, n (%)	4705 (24.63 %)	3498 (28.18 %)
6	First presentation depression, n (%)	11233 (58.79 %)	7453 (60.05 %)
7	Depression dominant, n (%)	9662 (50.57 %)	6457 (52.02 %)
8	Psychotic experiences, n (%)	4939 (25.85 %)	3806 (30.66 %)
9	Self-harm history, n (%)	2362 (12.36 %)	1822 (14.68 %)
10	Smoker, n (%)	13970 (73.12 %)	8796 (70.87 %)
11	Family history for bipolar disorder, n (%)	402 (2.10 %)	136 (1.10 %)
12	Family history for depression, n (%)	375 (1.96 %)	245 (1.97 %)
13	Family history for psychosis, n (%)	111 (0.58 %)	141 (1.14 %)
14	Overweight or obese, n (%)	6540 (34.23 %)	5038 (40.59 %)

Modification

Some values are negative. They should not be.

display(df.loc[df.symptom_to_exposure < 0, 'symptom_to_exposure'])
display(df.loc[df.age_first_diagnosis < 0, ['age_first_diagnosis', 'symptom_to_exposure']])
pd.set_option('display.max_rows', 100)
display(df.loc[(df.symptom_to_exposure>65) &  (df.symptom_to_exposure<80)])

We modify those illicit values

df.loc[df.symptom_to_exposure<0, 'symptom_to_exposure'] = 0
df.loc[df.age_first_diagnosis<0, 'age_first_diagnosis'] = np.nan
df.loc[df.symptom_to_exposure>=80, 'symptom_to_exposure'] = np.nan

Target Selection

target_multiclass = df.response2_1.replace(1, 2).fillna(1)
df['target_multiclass'] = target_multiclass

target_lithium2y = (df.exposure=='lithium') & (df.response2_1==1)
df['target_lithium2y'] = target_lithium2y.astype(int)

target_exposure = (df.exposure=='olanzapine')
df['target_exposure'] = target_exposure.astype(int)
targets = {
    'resp': 'response2_1',
    'exp': 'target_exposure',  # 0=lithium 1=olanzapine
    'multi': 'target_multiclass',
    'lithium2y': 'target_lithium2y'
}

for target in targets:
    print(df[targets[target]].value_counts())
    print()

0.0    14785
1.0    11848
Name: response2_1, dtype: int64

0    19106
1    12412
Name: target_exposure, dtype: int64

0.0    14785
2.0    11848
1.0     4885
Name: target_multiclass, dtype: int64

0    23501
1     8017
Name: target_lithium2y, dtype: int64

Feature Selection

We have a lot of "age" features. The ones that have few missing values are good candidates: let's keep only the ages with less than 5000 missing values.

age_columns = [col for col in df.columns if col.startswith('age_')]
with pd.option_context('display.max_rows', None, 'display.max_columns', None):
    age_df = df[age_columns].isna().sum().sort_values(ascending=True)
    display(age_df)
features_age = age_df[age_df < 5000].index.to_list()

age_first_exposure                    0
age_first_reg                         0
age_diagnosis                        21
age_first_diagnosis                  26
age_first                           143
age_smoke                           968
age_BP                             3102
age_eGFR                           5370
age_TSH                            9770
age_BMI                           10303
age_first_li                      10466
age_first_AP                      11000
age_depression                    11533
age_ethnicity                     12565
age_last_SSRI                     12704
age_first_SSRI                    12722
age_HDL                           14024
age_first_olan                    14456
age_transfer_out                  15942
age_LDL                           16581
age_ca                            16668
age_anxiety                       17770
age_first_MS                      17786
age_last_TCAs                     18235
age_first_TCAs                    18258
age_psychosis                     19271
age_last_other_ADs                19447
age_first_other_ADs               19451
age_sleep                         21187
age_mania                         21830
age_Chronic_pulmonary_disease     23645
age_self_harm                     24530
age_dermatitis                    24955
age_Diabetes_uncomplicated        25385
age_stress                        25915
age_asthma                        25949
age_Neurological_disorders        26065
age_T2DM                          26324
age_alcohol                       26479
age_death                         26786
age_hypothyroid                   27231
age_PD                            28117
age_CHD                           28361
age_relationship                  28484
age_migraine                      28782
age_other_substance_misuse        28977
age_Deficiency_anaemia            29025
age_cardiac_arrythmia             29594
age_Weight_loss                   29594
age_Fluid_electrolyte_disorder    29903
age_Diabetes_organ_damage         30053
age_RA                            30419
age_Congestive_heart_failure      30524
age_OCD                           30601
age_Peptic_ulcer                  30629
age_Peripheral_vascular           30642
age_cannabis                      30764
age_Liver_disease                 30781
age_Valvular_disease              30950
age_psych_FH                      31014
age_FH_NOS                        31052
age_Pulmonary_circulation         31056
age_hyperthyroid                  31077
age_FH_depression                 31108
age_Coagulopathy                  31208
age_adhd                          31217
age_FH_BPD                        31234
age_FH_psychosis                  31267
age_conduct                       31275
age_HIV_AIDS                      31289
age_FH_suicide                    31383
age_FH_substance                  31398
age_FH_anxiety                    31492
age_FH_LD                         31509
dtype: int64

We now remove features that do not make sense from a clinical perspective, possibly confusing the learning process

features_age.remove('age_BP')
features_age.remove('age_first')
features_age.remove('age_diagnosis')
features_age.remove('age_smoke')
print('age features:', features_age)

age features: ['age_first_exposure', 'age_first_reg', 'age_first_diagnosis']

We now select a good list of features, agnostic as well as informed

# list of agnostic features (to complement with age features)
features_agnostic = """adhd
FH_suicide
mania
psychosis
relationship
self_harm
sex
sleep
smoker
T2DM
OCD
migraine
hypothyroid
CHD
other_substance_misuse
cannabis
alcohol
depression
FH_anxiety
FH_any
FH_BPD
FH_depression
FH_LD
FH_psychosis
N_dep_b4
N_man_b4
anxiety
stress
hi_LDL
lo_HDL
weight
CKD3
symptom_to_exposure
dominant"""



# list of informed features:
features_informed = """age_first_exposure
age_first_diagnosis
symptom_to_exposure
psychosis
depression
mania
dominant
sex
FH_BPD
FH_depression
FH_psychosis
weight
self_harm
cannabis
anxiety
stress
sleep
other_substance_misuse
relationship
OCD
adhd
smoker
alcohol
FH_suicide
hi_LDL
lo_HDL
CKD3
T2DM
migraine
hypothyroid
CHD
FH_anxiety
FH_any
FH_LD"""



shaky_features="""cannabis
anxiety
stress
sleep
other_substance_misuse
relationship
OCD
adhd
smoker
alcohol
FH_suicide
hi_LDL
lo_HDL
CKD3
T2DM
migraine
hypothyroid
CHD
FH_anxiety
FH_any
FH_LD"""
shaky_features = [ feature.strip() for feature in shaky_features.split('\n') ]

features = {
    # informed
    '34': [ feature.strip() for feature in features_informed.split('\n') ],
    # agnostic
    '37' : [ feature.strip() for feature in features_agnostic.split("\n") ] + features_age
}
features.update({
    '13' : [feature for feature in features['34'] if feature not in shaky_features]
})

print(len(features['34']), 'informed features:', features['34'])
print(len(features['37']), 'agnostic features:', features['37'])
print(len(features['13']), 'important features:', features['13'])
print(len(shaky_features), 'shaky features:', shaky_features)

34 informed features: ['age_first_exposure', 'age_first_diagnosis', 'symptom_to_exposure', 'psychosis', 'depression', 'mania', 'dominant', 'sex', 'FH_BPD', 'FH_depression', 'FH_psychosis', 'weight', 'self_harm', 'cannabis', 'anxiety', 'stress', 'sleep', 'other_substance_misuse', 'relationship', 'OCD', 'adhd', 'smoker', 'alcohol', 'FH_suicide', 'hi_LDL', 'lo_HDL', 'CKD3', 'T2DM', 'migraine', 'hypothyroid', 'CHD', 'FH_anxiety', 'FH_any', 'FH_LD']
37 agnostic features: ['adhd', 'FH_suicide', 'mania', 'psychosis', 'relationship', 'self_harm', 'sex', 'sleep', 'smoker', 'T2DM', 'OCD', 'migraine', 'hypothyroid', 'CHD', 'other_substance_misuse', 'cannabis', 'alcohol', 'depression', 'FH_anxiety', 'FH_any', 'FH_BPD', 'FH_depression', 'FH_LD', 'FH_psychosis', 'N_dep_b4', 'N_man_b4', 'anxiety', 'stress', 'hi_LDL', 'lo_HDL', 'weight', 'CKD3', 'symptom_to_exposure', 'dominant', 'age_first_exposure', 'age_first_reg', 'age_first_diagnosis']
13 important features: ['age_first_exposure', 'age_first_diagnosis', 'symptom_to_exposure', 'psychosis', 'depression', 'mania', 'dominant', 'sex', 'FH_BPD', 'FH_depression', 'FH_psychosis', 'weight', 'self_harm']
21 shaky features: ['cannabis', 'anxiety', 'stress', 'sleep', 'other_substance_misuse', 'relationship', 'OCD', 'adhd', 'smoker', 'alcohol', 'FH_suicide', 'hi_LDL', 'lo_HDL', 'CKD3', 'T2DM', 'migraine', 'hypothyroid', 'CHD', 'FH_anxiety', 'FH_any', 'FH_LD']

print("Agnostic features that are not informed features:", list(set(features['37']) - set(features['34'])))
print("Informed features that are not agnostic features:", list(set(features['34']) - set(features['37'])))

Agnostic features that are not informed features: ['N_man_b4', 'N_dep_b4', 'age_first_reg']
Informed features that are not agnostic features: []

Preparation of the dataframes (samples=X and target=y)

Here we prepare the X dataframe (samples with their features) and the y dataframe (target to predict, for each sample). We also fix the types of a few features to make sure the algorithms will interpret them correctly.

from sklearn import preprocessing, metrics

def prepare(features, target):

    # We first remove the samples with N/A for any of the features or label
    df_withoutNA = df.dropna(subset=(features + [target]))
    df_features = df_withoutNA[features]

    # Now we encode all string values in the features into an int value (we might use One Hot encoding later?)
    le = preprocessing.LabelEncoder()
    to_encode = [key for key in dict(df_features.dtypes) if dict(df.dtypes)[key] not in ['float64', 'int64']]
    new_df_features = df_features.copy()
    new_df_features.update(df_features[to_encode].apply(le.fit_transform))
    new_df_features[to_encode] = new_df_features[to_encode].astype(np.int64)

    # Because lo_HDL has missing values, it was interpreted by pandas with
    # floats. Now that we removed the missing values, we can interpret as int
    if 'lo_HDL' in features:
        new_df_features.lo_HDL = new_df_features.lo_HDL.astype(np.int64)

    # Now we have our X and y
    return new_df_features, df_withoutNA[target]

X = dict()
y = dict()
for feature in features:
    for target in targets:
        X[feature + '_' + target], y[feature + '_' + target] = prepare(features[feature] + ['exposure'], targets[target])
        if target != 'exp':
            y[feature + '_' + target] = y[feature + '_' + target].astype(np.int64)
        print(feature + '_' + target, len(X[feature + '_' + target]), 'rows')

34_resp 26509 rows
34_exp 31369 rows
34_multi 31369 rows
34_lithium2y 31369 rows
37_resp 26509 rows
37_exp 31369 rows
37_multi 31369 rows
37_lithium2y 31369 rows
13_resp 26510 rows
13_exp 31370 rows
13_multi 31370 rows
13_lithium2y 31370 rows

from scipy.stats import pearsonr, spearmanr
import seaborn as sns

# The difference between the X dictionnary and the X_dict dictionnary is that we drop 'exposure' in X-dict. We don't want it as a feature in X-dict but we need it in X to know the exposure of the patient.
X_dict = dict()
y_dict = dict()
for key in X:
    X_dict.update({
        key: X[key].drop('exposure', axis=1)
    })
    X_dict.update({
        'num_' + key: X_dict[key].loc[:, X_dict[key].dtypes == np.float64],
        'cat_' + key: X_dict[key].loc[:, X_dict[key].dtypes == np.int64],
    })
    X_dict.update({
        'bin_' + key: X_dict['cat_' + key].loc[:, X_dict['cat_' + key].nunique() == 2]
    })

    y_dict.update({
        key: y[key],
        'num_' + key: y[key],
        'cat_' + key: y[key],
        'bin_' + key: y[key],
    })

    # When we predict exposure or lithium2y, it doesn't make sense to predict separately exposures
    if '_exp' not in key and '_lithium2y' not in key:
        X_dict.update({
            'lit_' + key: X[key].loc[X[key].exposure == 0].drop('exposure', axis=1),
            'ola_' + key: X[key].loc[X[key].exposure == 1].drop('exposure', axis=1),
        })
        y_dict.update({
            'lit_' + key : y[key][X[key].loc[X[key].exposure == 0].index],
            'ola_' + key : y[key][X[key].loc[X[key].exposure == 1].index],
        })


print(X_dict.keys())
print(y_dict.keys())

dict_keys(['34_resp', 'num_34_resp', 'cat_34_resp', 'bin_34_resp', 'lit_34_resp', 'ola_34_resp', '34_exp', 'num_34_exp', 'cat_34_exp', 'bin_34_exp', '34_multi', 'num_34_multi', 'cat_34_multi', 'bin_34_multi', 'lit_34_multi', 'ola_34_multi', '34_lithium2y', 'num_34_lithium2y', 'cat_34_lithium2y', 'bin_34_lithium2y', '37_resp', 'num_37_resp', 'cat_37_resp', 'bin_37_resp', 'lit_37_resp', 'ola_37_resp', '37_exp', 'num_37_exp', 'cat_37_exp', 'bin_37_exp', '37_multi', 'num_37_multi', 'cat_37_multi', 'bin_37_multi', 'lit_37_multi', 'ola_37_multi', '37_lithium2y', 'num_37_lithium2y', 'cat_37_lithium2y', 'bin_37_lithium2y', '13_resp', 'num_13_resp', 'cat_13_resp', 'bin_13_resp', 'lit_13_resp', 'ola_13_resp', '13_exp', 'num_13_exp', 'cat_13_exp', 'bin_13_exp', '13_multi', 'num_13_multi', 'cat_13_multi', 'bin_13_multi', 'lit_13_multi', 'ola_13_multi', '13_lithium2y', 'num_13_lithium2y', 'cat_13_lithium2y', 'bin_13_lithium2y'])
dict_keys(['34_resp', 'num_34_resp', 'cat_34_resp', 'bin_34_resp', 'lit_34_resp', 'ola_34_resp', '34_exp', 'num_34_exp', 'cat_34_exp', 'bin_34_exp', '34_multi', 'num_34_multi', 'cat_34_multi', 'bin_34_multi', 'lit_34_multi', 'ola_34_multi', '34_lithium2y', 'num_34_lithium2y', 'cat_34_lithium2y', 'bin_34_lithium2y', '37_resp', 'num_37_resp', 'cat_37_resp', 'bin_37_resp', 'lit_37_resp', 'ola_37_resp', '37_exp', 'num_37_exp', 'cat_37_exp', 'bin_37_exp', '37_multi', 'num_37_multi', 'cat_37_multi', 'bin_37_multi', 'lit_37_multi', 'ola_37_multi', '37_lithium2y', 'num_37_lithium2y', 'cat_37_lithium2y', 'bin_37_lithium2y', '13_resp', 'num_13_resp', 'cat_13_resp', 'bin_13_resp', 'lit_13_resp', 'ola_13_resp', '13_exp', 'num_13_exp', 'cat_13_exp', 'bin_13_exp', '13_multi', 'num_13_multi', 'cat_13_multi', 'bin_13_multi', 'lit_13_multi', 'ola_13_multi', '13_lithium2y', 'num_13_lithium2y', 'cat_13_lithium2y', 'bin_13_lithium2y'])

display(X_dict['cat_13_lithium2y'])

Feature Characterisation

Distribution and plots

%matplotlib inline
import seaborn as sns
sns.set_theme(style="white")
sns.set(font_scale=2)
from matplotlib.ticker import FormatStrFormatter

values = {0: df.age_first_exposure,
         1: df.age_first_diagnosis}

fig, ax = plt.subplots(ncols=2, nrows=1, sharey=True)
plt.subplots_adjust(hspace=0.6)
fig.set_size_inches(10, 6)
fig.tight_layout()
for i in values.keys():
    sns.distplot(values[i], kde=False, ax=ax[i], bins=range(100))
    # ax[i].set_yscale('log')
    ax[i].yaxis.set_major_formatter(FormatStrFormatter('%.0f'))
    ax[i].set_xlim((0,100))
display(df.age_first_diagnosis)

/Users/fehmi/GoogleDrive/sics/projects/ucl/lithium/venv/lib/python3.9/site-packages/seaborn/distributions.py:2619: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).
  warnings.warn(msg, FutureWarning)

1        80.722794
2        41.752224
3        56.533882
4        21.207392
5        82.948662
           ...    
38950    55.764545
38951    46.135525
38953    28.867899
38954    31.498974
38956    38.529774
Name: age_first_diagnosis, Length: 31518, dtype: float64

Now let's check the number of missing values for each feature, and look at their entropy

import scipy.stats as st

def entropy3(labels, base=None):
  vc = pd.Series(labels).value_counts(normalize=False, sort=False)
  base = 2 if base is None else base
  #return -(vc * np.log(vc)/np.log(base)).sum()
  return st.entropy(vc)

nan_entropy = pd.concat([df[features['37']].isna().sum(), df[features['37']].apply(entropy3, axis=0)], axis=1)
nan_entropy.columns = ['N/A', 'Entropy']
display(nan_entropy)
display(df[features['37']])

	N/A	Entropy
adhd	0	0.033217
FH_suicide	0	0.027631
mania	0	0.573333
psychosis	0	0.590539
relationship	0	0.390328
self_harm	0	0.391580
sex	0	0.679209
sleep	0	0.379125
smoker	0	1.066039
T2DM	0	0.175462
OCD	0	0.096136
migraine	0	0.203815
hypothyroid	0	0.201039
CHD	0	0.145867
other_substance_misuse	0	0.191553
cannabis	0	0.078576
alcohol	0	0.224948
depression	0	0.675798
FH_anxiety	0	0.010781
FH_any	0	0.233201
FH_BPD	0	0.086404
FH_depression	0	0.096757
FH_LD	0	0.002616
FH_psychosis	0	0.046572
N_dep_b4	0	1.995139
N_man_b4	0	0.893913
anxiety	0	0.547383
stress	0	0.302052
hi_LDL	0	0.228529
lo_HDL	1	0.119001
weight	0	1.014700
CKD3	0	0.094266
symptom_to_exposure	143	8.074818
dominant	0	0.988668
age_first_exposure	0	9.559759
age_first_reg	0	9.567249
age_first_diagnosis	26	8.743020

	adhd	FH_suicide	mania	psychosis	relationship	self_harm	sex	sleep	smoker	T2DM	OCD	migraine	hypothyroid	CHD	other_substance_misuse	cannabis	alcohol	depression	FH_anxiety	FH_any	FH_BPD	FH_depression	FH_LD	FH_psychosis	N_dep_b4	N_man_b4	anxiety	stress	hi_LDL	lo_HDL	weight	CKD3	symptom_to_exposure	dominant	age_first_exposure	age_first_reg	age_first_diagnosis
1	0	0	0	1	0	0	female	0	ex smoker	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0.0	healthy weight	0	0.000000	unclear	69.434631	64.391781	80.722794
2	0	0	0	0	0	1	female	1	ex smoker	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	1	0	1	0	1	0.0	healthy weight	0	24.128679	depression	43.627651	27.057534	41.752224
3	0	0	1	0	0	0	female	0	current smoker	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0.0	overweight	0	0.000000	mania	56.533882	50.843836	56.533882
4	0	0	1	0	0	1	female	1	ex smoker	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	12	1	0	0	0	0.0	healthy weight	0	10.973306	depression	26.524298	22.583562	21.207392
5	0	0	0	1	0	0	female	0	current smoker	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	2	0	0	0	0	0.0	healthy weight	0	13.349760	depression	72.032852	67.394521	82.948662
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
38950	0	0	0	0	1	1	female	0	current smoker	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0.0	healthy weight	1	18.521561	unclear	74.286102	45.613699	55.764545
38951	0	0	0	1	0	0	female	0	never smoker	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	1	0	1	0	0	0.0	healthy weight	0	7.293634	depression	43.791924	44.260274	46.135525
38953	0	0	0	0	1	0	female	1	current smoker	0	0	0	0	0	1	1	0	1	0	1	0	0	0	0	13	0	1	1	0	0.0	overweight	0	11.707050	depression	37.223820	45.471233	28.867899
38954	0	0	1	0	0	0	female	0	ex smoker	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	1	0	0	0.0	healthy weight	0	16.922655	mania	48.421631	39.983562	31.498974
38956	0	0	0	1	0	0	female	0	current smoker	0	0	0	1	0	0	0	0	1	0	0	0	0	0	0	2	0	0	0	0	0.0	healthy weight	0	14.631075	depression	39.129364	30.767123	38.529774

31518 rows × 37 columns

key = '34_exp'
import itertools
variables_pearson = X_dict['num_' + key].columns
for variables in itertools.combinations(variables_pearson, 2):
    plt.figure(figsize=(16,9))
    sns.regplot(x = variables[0], y = variables[1], data = X_dict['num_'+key],
            line_kws={"color": "red"})

display(X_dict['num_'+ key])

	age_first_exposure	age_first_diagnosis	symptom_to_exposure
1	69.434631	80.722794	0.000000
2	43.627651	41.752224	24.128679
3	56.533882	56.533882	0.000000
4	26.524298	21.207392	10.973306
5	72.032852	82.948662	13.349760
...	...	...	...
38950	74.286102	55.764545	18.521561
38951	43.791924	46.135525	7.293634
38953	37.223820	28.867899	11.707050
38954	48.421631	31.498974	16.922655
38956	39.129364	38.529774	14.631075

31369 rows × 3 columns

Correlation coefficients

With target

from pyitlib import discrete_random_variable as drv
from scipy.spatial import distance

def similarity(X,Y):
    sim = ((X*Y)+((1-X)*(1-Y)))/len(X)
    return(sim.sum())

def coeff(X_dict, y, key, feature):
    print(len(features[feature]), 'features')
    df_coeff = pd.DataFrame(index=features[feature])

    # Now we calculate both for each numerical feature
    for feat in features[feature]:
        # We calculate correlation coeffs only for numerical and binary categorical
        if (feat in X_dict['num_' + key].columns):
            df_coeff.loc[feat, 'pearson_r'], df_coeff.loc[feat, 'pearson_p'] = pearsonr(X[key][feat], y[key])
            df_coeff.loc[feat, 'spearman_r'], df_coeff.loc[feat, 'spearman_p'] = spearmanr(X[key][feat], y[key])
        # We calculate conditional entropy for only categorical (including non binary)
        if feat in X_dict['cat_'+ key].columns:
            df_coeff.loc[feat, 'cond_entropy'] = drv.entropy_conditional([int(x) for x in X[key][feat]], [int(x) for x in y[key]], base=np.e)
        # For binary variables the Jaccard similarity is more approriate
        if (feat in X_dict['bin_' + key].columns) and (y[key].dtype != np.float64):
            df_coeff.loc[feat, 'jaccard'] = 1-distance.jaccard(X[key][feat], y[key])
            df_coeff.loc[feat, 'SMC'] = similarity(X[key][feat], y[key])

    # We display the sorted absolute values only when p-value < 0.01
    display(df_coeff
            .reindex(df_coeff.pearson_r.abs().sort_values(ascending=False).index))
    return df_coeff

For binary features:

• Jaccard coefficient only counts how many "1" are similar (eg, used a lot when analysing shopping similarities between shoppers).
  • 0 = feature and target are never both "1" for each patient
  • 1 = feature and target are always both "1" together for each patient
• SMC (Simple Similarity Matching Coeff) counts both "1" and "0"
  • 0 = feature and target are never the same
  • 1 = feature and target are always the same

feature = '34'
for target in targets:
    key = feature + '_' + target
    print(10*'_' + target + 10*'_')
    df_coeff = coeff(X_dict, y, key, feature)
    print()

__________resp__________
34 features

	pearson_r	pearson_p	spearman_r	spearman_p	cond_entropy	jaccard	SMC
age_first_exposure	0.193293	1.693068e-221	0.195700	4.192704e-227	NaN	NaN	NaN
age_first_diagnosis	0.131446	1.780416e-102	0.130364	8.174834e-101	NaN	NaN	NaN
symptom_to_exposure	0.122443	4.498120e-89	0.119015	3.168104e-84	NaN	NaN	NaN
psychosis	NaN	NaN	NaN	NaN	0.589143	0.209603	0.529292
depression	NaN	NaN	NaN	NaN	0.676162	0.345302	0.495832
mania	NaN	NaN	NaN	NaN	0.574144	0.197438	0.527255
dominant	NaN	NaN	NaN	NaN	0.989417	NaN	NaN
sex	NaN	NaN	NaN	NaN	0.679791	0.274790	0.509186
FH_BPD	NaN	NaN	NaN	NaN	0.086441	0.015442	0.552642
FH_depression	NaN	NaN	NaN	NaN	0.095472	0.017635	0.552416
FH_psychosis	NaN	NaN	NaN	NaN	0.044860	0.006384	0.553774
weight	NaN	NaN	NaN	NaN	1.014475	NaN	NaN
self_harm	NaN	NaN	NaN	NaN	0.391276	0.108984	0.536459
cannabis	NaN	NaN	NaN	NaN	0.076336	0.011893	0.551813
anxiety	NaN	NaN	NaN	NaN	0.548561	0.172467	0.518352
stress	NaN	NaN	NaN	NaN	0.300254	0.064856	0.531140
sleep	NaN	NaN	NaN	NaN	0.379208	0.104652	0.537516
other_substance_misuse	NaN	NaN	NaN	NaN	0.192186	0.044094	0.549398
relationship	NaN	NaN	NaN	NaN	0.392561	0.113042	0.539741
OCD	NaN	NaN	NaN	NaN	0.098012	0.019634	0.553586
adhd	NaN	NaN	NaN	NaN	0.033642	0.004212	0.554038
smoker	NaN	NaN	NaN	NaN	1.065933	NaN	NaN
alcohol	NaN	NaN	NaN	NaN	0.227904	0.049326	0.542684
FH_suicide	NaN	NaN	NaN	NaN	0.026897	0.004309	0.555396
hi_LDL	NaN	NaN	NaN	NaN	0.229586	0.055144	0.547550
lo_HDL	NaN	NaN	NaN	NaN	0.119419	0.025348	0.553246
CKD3	NaN	NaN	NaN	NaN	0.092342	0.019442	0.554793
T2DM	NaN	NaN	NaN	NaN	0.175593	0.049850	0.559244
migraine	NaN	NaN	NaN	NaN	0.203305	0.044653	0.546418
hypothyroid	NaN	NaN	NaN	NaN	0.201963	0.057541	0.558226
CHD	NaN	NaN	NaN	NaN	0.144386	0.036291	0.556226
FH_anxiety	NaN	NaN	NaN	NaN	0.009570	0.001187	0.555509
FH_any	NaN	NaN	NaN	NaN	0.231687	0.052400	0.544306
FH_LD	NaN	NaN	NaN	NaN	0.002625	0.000594	0.555924

__________exp__________
34 features

	pearson_r	pearson_p	spearman_r	spearman_p	cond_entropy	jaccard	SMC
age_first_exposure	-0.106543	7.323420e-80	-0.108190	2.643016e-82	NaN	NaN	NaN
symptom_to_exposure	-0.075266	1.201673e-40	-0.082721	9.258229e-49	NaN	NaN	NaN
age_first_diagnosis	-0.059584	4.470683e-26	-0.054850	2.440734e-22	NaN	NaN	NaN
psychosis	NaN	NaN	NaN	NaN	0.589280	0.219806	0.570021
depression	NaN	NaN	NaN	NaN	0.676135	0.315729	0.487009
mania	NaN	NaN	NaN	NaN	0.572370	0.204563	0.567630
dominant	NaN	NaN	NaN	NaN	0.988610	NaN	NaN
sex	NaN	NaN	NaN	NaN	0.678129	0.278029	0.541394
FH_BPD	NaN	NaN	NaN	NaN	0.085809	0.010562	0.596831
FH_depression	NaN	NaN	NaN	NaN	0.096748	0.019133	0.601231
FH_psychosis	NaN	NaN	NaN	NaN	0.046308	0.011288	0.606299
weight	NaN	NaN	NaN	NaN	1.010127	NaN	NaN
self_harm	NaN	NaN	NaN	NaN	0.391315	0.123489	0.588415
cannabis	NaN	NaN	NaN	NaN	0.075429	0.027083	0.611782
anxiety	NaN	NaN	NaN	NaN	0.541934	0.224798	0.600179
stress	NaN	NaN	NaN	NaN	0.293544	0.126974	0.624534
sleep	NaN	NaN	NaN	NaN	0.373184	0.150662	0.615416
other_substance_misuse	NaN	NaN	NaN	NaN	0.185501	0.074975	0.619274
relationship	NaN	NaN	NaN	NaN	0.388881	0.092250	0.562402
OCD	NaN	NaN	NaN	NaN	0.095491	0.025091	0.606108
adhd	NaN	NaN	NaN	NaN	0.033051	0.007548	0.605980
smoker	NaN	NaN	NaN	NaN	1.060558	NaN	NaN
alcohol	NaN	NaN	NaN	NaN	0.222606	0.075436	0.609678
FH_suicide	NaN	NaN	NaN	NaN	0.027568	0.004173	0.604386
hi_LDL	NaN	NaN	NaN	NaN	0.223190	0.087069	0.617616
lo_HDL	NaN	NaN	NaN	NaN	0.117458	0.036801	0.609519
CKD3	NaN	NaN	NaN	NaN	0.093997	0.023179	0.605024
T2DM	NaN	NaN	NaN	NaN	0.175785	0.039021	0.595684
migraine	NaN	NaN	NaN	NaN	0.203515	0.056255	0.601039
hypothyroid	NaN	NaN	NaN	NaN	0.200442	0.037254	0.586439
CHD	NaN	NaN	NaN	NaN	0.145779	0.035962	0.601772
FH_anxiety	NaN	NaN	NaN	NaN	0.010400	0.001532	0.605183
FH_any	NaN	NaN	NaN	NaN	0.233395	0.057539	0.592719
FH_LD	NaN	NaN	NaN	NaN	0.002539	0.000565	0.605502

__________multi__________
34 features

	pearson_r	pearson_p	spearman_r	spearman_p	cond_entropy	jaccard	SMC
age_first_exposure	0.176324	1.914504e-217	0.177382	4.462611e-220	NaN	NaN	NaN
age_first_diagnosis	0.120163	3.210233e-101	0.118399	2.643382e-98	NaN	NaN	NaN
symptom_to_exposure	0.112088	3.062966e-88	0.107891	7.372539e-82	NaN	NaN	NaN
psychosis	NaN	NaN	NaN	NaN	0.590599	0.067322	0.327106
depression	NaN	NaN	NaN	NaN	0.676158	0.113832	0.584686
mania	NaN	NaN	NaN	NaN	0.573145	0.060513	0.306162
dominant	NaN	NaN	NaN	NaN	0.989188	NaN	NaN
sex	NaN	NaN	NaN	NaN	0.679341	0.087408	0.432688
FH_BPD	NaN	NaN	NaN	NaN	0.086569	0.004969	0.106092
FH_depression	NaN	NaN	NaN	NaN	0.096706	0.006140	0.108260
FH_psychosis	NaN	NaN	NaN	NaN	0.046645	0.002923	0.098919
weight	NaN	NaN	NaN	NaN	1.014602	NaN	NaN
self_harm	NaN	NaN	NaN	NaN	0.391790	0.035012	0.194523
cannabis	NaN	NaN	NaN	NaN	0.078415	0.005153	0.102745
anxiety	NaN	NaN	NaN	NaN	0.547361	0.055399	0.268067
stress	NaN	NaN	NaN	NaN	0.301458	0.024737	0.142657
sleep	NaN	NaN	NaN	NaN	0.379355	0.033202	0.189773
other_substance_misuse	NaN	NaN	NaN	NaN	0.191935	0.013252	0.131276
relationship	NaN	NaN	NaN	NaN	0.389897	0.032445	0.199050
OCD	NaN	NaN	NaN	NaN	0.096090	0.004843	0.110077
adhd	NaN	NaN	NaN	NaN	0.033256	0.001434	0.096688
smoker	NaN	NaN	NaN	NaN	1.065385	NaN	NaN
alcohol	NaN	NaN	NaN	NaN	0.224836	0.014706	0.131499
FH_suicide	NaN	NaN	NaN	NaN	0.027556	0.001437	0.097899
hi_LDL	NaN	NaN	NaN	NaN	0.228775	0.016350	0.141031
lo_HDL	NaN	NaN	NaN	NaN	0.119321	0.007289	0.115656
CKD3	NaN	NaN	NaN	NaN	0.094304	0.006274	0.111671
T2DM	NaN	NaN	NaN	NaN	0.175073	0.011715	0.142625
migraine	NaN	NaN	NaN	NaN	0.203864	0.014731	0.130320
hypothyroid	NaN	NaN	NaN	NaN	0.200578	0.013537	0.149319
CHD	NaN	NaN	NaN	NaN	0.145722	0.010078	0.128311
FH_anxiety	NaN	NaN	NaN	NaN	0.010378	0.000600	0.095190
FH_any	NaN	NaN	NaN	NaN	0.233232	0.018060	0.137237
FH_LD	NaN	NaN	NaN	NaN	0.002521	0.000060	0.094807

__________lithium2y__________
34 features

	pearson_r	pearson_p	spearman_r	spearman_p	cond_entropy	jaccard	SMC
age_first_exposure	0.154052	7.457636e-166	0.157628	1.206823e-173	NaN	NaN	NaN
age_first_diagnosis	0.104651	4.215269e-77	0.101220	3.181159e-72	NaN	NaN	NaN
symptom_to_exposure	0.103478	2.051218e-75	0.101697	6.819342e-73	NaN	NaN	NaN
psychosis	NaN	NaN	NaN	NaN	0.590447	0.142651	0.601294
depression	NaN	NaN	NaN	NaN	0.676061	0.222709	0.462495
mania	NaN	NaN	NaN	NaN	0.572964	0.137856	0.610635
dominant	NaN	NaN	NaN	NaN	0.988927	NaN	NaN
sex	NaN	NaN	NaN	NaN	0.679087	0.176928	0.530779
FH_BPD	NaN	NaN	NaN	NaN	0.086570	0.018090	0.738723
FH_depression	NaN	NaN	NaN	NaN	0.096748	0.018281	0.736364
FH_psychosis	NaN	NaN	NaN	NaN	0.046481	0.004525	0.740540
weight	NaN	NaN	NaN	NaN	1.014417	NaN	NaN
self_harm	NaN	NaN	NaN	NaN	0.391424	0.082263	0.672097
cannabis	NaN	NaN	NaN	NaN	0.077634	0.006322	0.734419
anxiety	NaN	NaN	NaN	NaN	0.546391	0.115886	0.611113
stress	NaN	NaN	NaN	NaN	0.299668	0.041727	0.683732
sleep	NaN	NaN	NaN	NaN	0.378212	0.070755	0.670088
other_substance_misuse	NaN	NaN	NaN	NaN	0.190062	0.022924	0.711945
relationship	NaN	NaN	NaN	NaN	0.389486	0.109298	0.690331
OCD	NaN	NaN	NaN	NaN	0.095951	0.014316	0.734419
adhd	NaN	NaN	NaN	NaN	0.033194	0.003209	0.742516
smoker	NaN	NaN	NaN	NaN	1.062893	NaN	NaN
alcohol	NaN	NaN	NaN	NaN	0.223868	0.032913	0.706812
FH_suicide	NaN	NaN	NaN	NaN	0.027538	0.005091	0.744557
hi_LDL	NaN	NaN	NaN	NaN	0.227714	0.035373	0.707036
lo_HDL	NaN	NaN	NaN	NaN	0.119160	0.019302	0.731136
CKD3	NaN	NaN	NaN	NaN	0.094344	0.016388	0.735950
T2DM	NaN	NaN	NaN	NaN	0.175457	0.046268	0.729924
migraine	NaN	NaN	NaN	NaN	0.203862	0.039913	0.717811
hypothyroid	NaN	NaN	NaN	NaN	0.200551	0.056950	0.728139
CHD	NaN	NaN	NaN	NaN	0.145969	0.029967	0.729638
FH_anxiety	NaN	NaN	NaN	NaN	0.010389	0.001001	0.745354
FH_any	NaN	NaN	NaN	NaN	0.233339	0.048752	0.713252
FH_LD	NaN	NaN	NaN	NaN	0.002609	0.000125	0.745991

Pairwise

from pandas import Series, DataFrame
import pandas as pd
import numpy as np
from collections import Counter
import os
from datetime import date
from sklearn.feature_selection import chi2
from scipy import stats
import seaborn as sns
import matplotlib.pylab as plt
from numpy import percentile
from sklearn.feature_selection import SelectKBest

def chisquare(Y):
    X = Y.loc[:, Y.dtypes == np.int64]
    column_names=X.columns
    chisqmatrix=pd.DataFrame(X,columns=column_names,index=column_names)

    outercnt=0
    innercnt=0
    for icol in column_names:

        for jcol in column_names:

           mycrosstab = pd.crosstab(X[icol], X[jcol])
           stat, p, dof, expected=stats.chi2_contingency(mycrosstab)
           chisqmatrix.iloc[outercnt,innercnt] = round(p,3)
           cntexpected = expected[expected<5].size
           perexpected = ((expected.size-cntexpected)/expected.size)*100

           if perexpected < 20:
                chisqmatrix.iloc[outercnt,innercnt] = 2
           if icol==jcol:
               chisqmatrix.iloc[outercnt,innercnt]=0.00
           innercnt = innercnt + 1
        outercnt = outercnt + 1
        innercnt = 0

    return chisqmatrix

df_corr_pearson = X_dict['num_'+ key].corr(method='pearson', min_periods=1)
df_corr_spearman = X_dict['num_' + key].corr(method='spearman', min_periods=1)
df_corr_chisquare = chisquare(X_dict['cat_' + key])

print(X_dict[key].dtypes)

age_first_exposure        float64
age_first_diagnosis       float64
symptom_to_exposure       float64
psychosis                   int64
depression                  int64
mania                       int64
dominant                    int64
sex                         int64
FH_BPD                      int64
FH_depression               int64
FH_psychosis                int64
weight                      int64
self_harm                   int64
cannabis                    int64
anxiety                     int64
stress                      int64
sleep                       int64
other_substance_misuse      int64
relationship                int64
OCD                         int64
adhd                        int64
smoker                      int64
alcohol                     int64
FH_suicide                  int64
hi_LDL                      int64
lo_HDL                      int64
CKD3                        int64
T2DM                        int64
migraine                    int64
hypothyroid                 int64
CHD                         int64
FH_anxiety                  int64
FH_any                      int64
FH_LD                       int64
dtype: object

import matplotlib.pyplot as plt
%matplotlib inline
sns.set_theme(style="white")
sns.set(font_scale=2)
def plot_matrix_corr(df_corr, fontsize=16, xsize=9, ysize=9):
    cmap = sns.diverging_palette(230, 20, as_cmap=True)
    plt.figure(figsize=(xsize,ysize))
    sns.heatmap(df_corr, xticklabels = df_corr.columns, yticklabels = df_corr.columns, annot=True,
                linewidths=0.5, cmap = cmap, fmt='.3f',
                annot_kws={
                'fontsize': fontsize,
                'fontweight': 'bold',
                'fontfamily': 'serif'}
                )

plot_matrix_corr(df_corr_pearson, 32, 16, 12)
plot_matrix_corr(df_corr_spearman, 32, 16, 12)

sns.set_theme(style="white")
plot_matrix_corr(df_corr_chisquare, 8, 22, 14)

About Condition Entropy.

Here, $X$ is the response and $Y$ is the feature being considered. Two properties:

1. Conditional entropy equals zero
   $ {H} (Y|X)=0 $ if and only if the value of $Y$ is completely determined by the value of $X$.
2. Conditional entropy of independent random variables
   Conversely, ${H} (Y|X)= {H} (Y)$ if and only if $Y$ and $X$ are independent random variables.

F-Test

import scipy.stats as stats

def anova(X_dict, key, target):
    anova_df = X_dict['num_' + key].merge(df[target], left_index=True, right_index=True)
    results_list = list()
    for feature in X_dict['num_' + key].keys():
        res = stats.f_oneway(anova_df.loc[anova_df[target]==0.0, feature].values,
                         anova_df.loc[anova_df[target]==1.0, feature].values)
        results_list = results_list + [[feature, res.statistic, res.pvalue]]
    return (pd.DataFrame(results_list, columns=['feature', 'F-Statistic', 'p-value']))

feature = '34'
for target in targets:
    key = feature + '_' + target
    print('key:', key)
    display(anova(X_dict, key, targets[target]))

key: 34_resp

	feature	F-Statistic	p-value
0	age_first_exposure	1028.801036	1.693068e-221
1	age_first_diagnosis	466.040468	1.780416e-102
2	symptom_to_exposure	403.450918	4.498120e-89

key: 34_exp

	feature	F-Statistic	p-value
0	age_first_exposure	360.151148	7.323420e-80
1	age_first_diagnosis	111.756084	4.470683e-26
2	symptom_to_exposure	178.703821	1.201673e-40

key: 34_multi

	feature	F-Statistic	p-value
0	age_first_exposure	46.909192	7.655806e-12
1	age_first_diagnosis	27.350178	1.714934e-07
2	symptom_to_exposure	13.520850	2.365581e-04

key: 34_lithium2y

	feature	F-Statistic	p-value
0	age_first_exposure	762.493116	7.457636e-166
1	age_first_diagnosis	347.333055	4.215269e-77
2	symptom_to_exposure	339.500883	2.051218e-75

T-Test

import scipy.stats as stats

for target in targets:
    key = feature + '_' + target
    print('key:', key)
    sample = 1000
    ttest_df = X_dict['num_' + key].merge(df[targets[target]], left_index=True, right_index=True)
    results_list = list()
    for feat in X_dict['num_' + key].keys():
        res = stats.ttest_ind(ttest_df.loc[ttest_df[targets[target]]==0.0, feat].head(sample),
                              ttest_df.loc[ttest_df[targets[target]]==1.0, feat].head(sample),
                              equal_var = True)
        results_list = results_list + [[feat, res.statistic, res.pvalue]]

    ttest_results = pd.DataFrame(results_list, columns=['feature', 'T-Statistic', 'p-value'])
    display(ttest_results)

key: 34_resp

	feature	T-Statistic	p-value
0	age_first_exposure	-8.560086	2.212296e-17
1	age_first_diagnosis	-4.615607	4.168509e-06
2	symptom_to_exposure	-6.759800	1.807514e-11

key: 34_exp

	feature	T-Statistic	p-value
0	age_first_exposure	5.359331	9.317705e-08
1	age_first_diagnosis	3.378131	7.438096e-04
2	symptom_to_exposure	3.309470	9.513413e-04

key: 34_multi

	feature	T-Statistic	p-value
0	age_first_exposure	-3.410012	0.000663
1	age_first_diagnosis	-2.348080	0.018967
2	symptom_to_exposure	-2.479474	0.013240

key: 34_lithium2y

	feature	T-Statistic	p-value
0	age_first_exposure	-7.933297	3.521460e-15
1	age_first_diagnosis	-3.810229	1.430358e-04
2	symptom_to_exposure	-6.630426	4.294092e-11

Dataset Splitting

We split by time (exposure start). The 80% first patients as training set, the 20% last patients as test set.

x_train = dict()
y_train = dict()
x_test = dict()
y_test = dict()

format='%d%b%Y'
df['exposure_start'] = pd.to_datetime(df['exposure_start'], format=format)
for feature_set in X_dict:
    print(feature_set)
    X_sorted = X_dict[feature_set].merge(df['exposure_start'], left_index=True, right_index=True).sort_values(by = 'exposure_start')
    y_sorted = y_dict[feature_set][X_sorted.index]
    limit = int(len(X_sorted)*.8) # We keep 80% for the training set
    train_index = X_sorted.index[:limit]
    test_index = X_sorted.index[limit:]
    x_train[feature_set] = X_dict[feature_set][X_dict[feature_set].index.isin(train_index)]
    y_train[feature_set] = y_dict[feature_set][y_dict[feature_set].index.isin(train_index)]
    x_test[feature_set] = X_dict[feature_set][X_dict[feature_set].index.isin(test_index)]
    y_test[feature_set] = y_dict[feature_set][y_dict[feature_set].index.isin(test_index)]

34_resp
num_34_resp
cat_34_resp
bin_34_resp
lit_34_resp
ola_34_resp
34_exp
num_34_exp
cat_34_exp
bin_34_exp
34_multi
num_34_multi
cat_34_multi
bin_34_multi
lit_34_multi
ola_34_multi
34_lithium2y
num_34_lithium2y
cat_34_lithium2y
bin_34_lithium2y
37_resp
num_37_resp
cat_37_resp
bin_37_resp
lit_37_resp
ola_37_resp
37_exp
num_37_exp
cat_37_exp
bin_37_exp
37_multi
num_37_multi
cat_37_multi
bin_37_multi
lit_37_multi
ola_37_multi
37_lithium2y
num_37_lithium2y
cat_37_lithium2y
bin_37_lithium2y
13_resp
num_13_resp
cat_13_resp
bin_13_resp
lit_13_resp
ola_13_resp
13_exp
num_13_exp
cat_13_exp
bin_13_exp
13_multi
num_13_multi
cat_13_multi
bin_13_multi
lit_13_multi
ola_13_multi
13_lithium2y
num_13_lithium2y
cat_13_lithium2y
bin_13_lithium2y

Evaluation function

Balanced Accuracy is the same as macro-averaged Recall, see e.g.: https://github.com/EpistasisLab/tpot/issues/108 Also confirmed here: https://amueller.github.io/aml/04-model-evaluation/10-evaluation-metrics.html

from sklearn.metrics import roc_auc_score

def results(clf, X1, y, v):
    X = X1
    if isinstance(X1, pd.Series):
        X = X1.to_frame()

    y_pred = clf.predict(X)
    y_score = clf.predict_proba(X)[:, 1]
    target_names = ['No Response', 'Response']
    multiclass = 'raise'

    if '_multi' in v:
        y_score = clf.predict_proba(X)
        multiclass = 'ovr'
        target_names = ['No Response', 'Equivocal', 'Response']
    if '_exp' in v:
        target_names = ['Lithium', 'Olanzapine']
    if '_lithium2y' in v:
        target_names = ['other', 'lithium2y']

    #print(classification_report(y, y_pred, target_names=target_names))
    result = classification_report(y, y_pred, target_names=target_names, output_dict=True)
    # Compute confusion matrix
    cm = confusion_matrix(y, y_pred)
    #print(cm)

    print('Balanced Accuracy:', result['macro avg']['recall'])

    # For average and multi_class parameters, check doc:
    # https://scikit-learn.org/stable/modules/generated/sklearn.metrics.roc_auc_score.html#sklearn.metrics.roc_auc_score
    try:
        roc_auc_score_val = roc_auc_score(y, y_score, average='weighted', multi_class=multiclass)
    except:
        # 34_multi fails with Naive Bayes because MixedNB predict_proba() seems to return proba values that don't sum up to 1...
        roc_auc_score_val = None
    result['roc_auc'] = roc_auc_score_val
    print('ROC_AUC score:', roc_auc_score_val)

    # Show confusion matrix in a separate window
    color_map = plt.cm.get_cmap('Blues')
    plt.matshow(cm, cmap=color_map)
    plt.title('Confusion matrix\n')
    plt.colorbar()
    plt.ylabel('True label')
    plt.xlabel('Predicted label')
    plt.show()

    return(result)

import warnings

def evaluate(clf, x_test, y_test):
    all_results = pd.DataFrame(columns=['features',
                                        'balanced accuracy',
                                        'accuracy',
                                        'roc_auc',
                                    'f1 (response)',
                                    'f1 (equivocal)',
                                    'f1 (no response)',
                                    'f1 (lithium)',
                                    'f1 (olanzapine)',
                                    'f1 (lithium > 2y)',
                                    'f1 (other)',
                                    'f1 (weighted avg)'])
    for v in clf:
        print(40*'_' + v + 40*'_')
        v2 = v.replace('_balanced_accuracy', '').replace('_accuracy', '').replace('_f1_weighted', '')
        with warnings.catch_warnings():
            warnings.simplefilter("ignore")
            result = results(clf[v], x_test[v2], y_test[v2], v)
        result_dict = {'features': v,
                        'accuracy': result['accuracy'],
                        'roc_auc': result['roc_auc'],
                       'balanced accuracy': result['macro avg']['recall'],
                        'f1 (weighted avg)': result['weighted avg']['f1-score']
                       }
        if '_exp' in v:
            result_dict.update({
                'f1 (lithium)': result['Lithium']['f1-score'],
                'f1 (olanzapine)': result['Olanzapine']['f1-score'],

            })
        elif '_multi' in v:
            result_dict.update({
                'f1 (no response)': result['No Response']['f1-score'],
                'f1 (equivocal)': result['Equivocal']['f1-score'],
                'f1 (response)': result['Response']['f1-score']
            })
        elif '_lithium2y' in v:
            result_dict.update({
                'f1 (other)': result['other']['f1-score'],
                'f1 (lithium > 2y)': result['lithium2y']['f1-score']
            })
        else:
            result_dict.update({
                'f1 (no response)': result['No Response']['f1-score'],
                'f1 (response)': result['Response']['f1-score']
            })

        all_results = all_results.append(result_dict,
                                         ignore_index=True)
        print(83*'_')
        print('\n\n\n')
    return all_results

Logistic Regression with Elastic Net regularisation

from sklearn.utils._testing import ignore_warnings
from sklearn.exceptions import ConvergenceWarning
from sklearn.linear_model import LogisticRegressionCV

@ignore_warnings(category=ConvergenceWarning) # max_iter default value (=100?) triggers this warning
def run(X1, y, l1, l2):
    X = X1
    if isinstance(X1, pd.Series):
        X = X1.to_frame()
    return LogisticRegressionCV(penalty='elasticnet',
                                l1_ratios=[l1, l2],
                                cv=5,
                                solver='saga',
                                scoring='balanced_accuracy',
                                n_jobs=4).fit(X, y)

clf_multi_dict = dict()

# Elastic net regularisation for larger feature sets
l1 = .8
l2 = .2

for v in [f for f in X_dict if len(X_dict[f].columns) > 5]:
    print(v)
    clf_multi_dict[v] = run(x_train[v], y_train[v], l1, l2) # a lot of features (more than 5)

for v in [f for f in X_dict if len(X_dict[f].columns) <= 5]:
    print(v)
    clf_multi_dict[v] = run(x_train[v], y_train[v], l2, l1) # NOT a lot of features (5 or less)

Results (interpret ROC_AUC with care because of class imbalance)

all_results = evaluate(clf_multi_dict, x_test, y_test)

________________________________________34_resp________________________________________
Balanced Accuracy: 0.5807004496516058
ROC_AUC score: 0.6445436122012334

___________________________________________________________________________________




________________________________________cat_34_resp________________________________________
Balanced Accuracy: 0.5206806844315283
ROC_AUC score: 0.5627429177011705

___________________________________________________________________________________




________________________________________bin_34_resp________________________________________
Balanced Accuracy: 0.517216479216485
ROC_AUC score: 0.5665701765426024

___________________________________________________________________________________




________________________________________lit_34_resp________________________________________
Balanced Accuracy: 0.576376784572234
ROC_AUC score: 0.6225357816631882

___________________________________________________________________________________




________________________________________ola_34_resp________________________________________
Balanced Accuracy: 0.5623262698430254
ROC_AUC score: 0.6852710006119724

___________________________________________________________________________________




________________________________________34_exp________________________________________
Balanced Accuracy: 0.5567292093418088
ROC_AUC score: 0.597591654042374

___________________________________________________________________________________




________________________________________cat_34_exp________________________________________
Balanced Accuracy: 0.5528329793498448
ROC_AUC score: 0.585100734306033

___________________________________________________________________________________




________________________________________bin_34_exp________________________________________
Balanced Accuracy: 0.5471165025089209
ROC_AUC score: 0.5855459922694032

___________________________________________________________________________________




________________________________________34_multi________________________________________
Balanced Accuracy: 0.3857732604055686
ROC_AUC score: 0.6075739122604424

___________________________________________________________________________________




________________________________________cat_34_multi________________________________________
Balanced Accuracy: 0.34700607363579766
ROC_AUC score: 0.5446438972107863

___________________________________________________________________________________




________________________________________bin_34_multi________________________________________
Balanced Accuracy: 0.3461819988708128
ROC_AUC score: 0.5478197984652663

___________________________________________________________________________________




________________________________________lit_34_multi________________________________________
Balanced Accuracy: 0.38412353361957013
ROC_AUC score: 0.5939621512078251

___________________________________________________________________________________




________________________________________ola_34_multi________________________________________
Balanced Accuracy: 0.37496773392901467
ROC_AUC score: 0.6348224798138037

___________________________________________________________________________________




________________________________________34_lithium2y________________________________________
Balanced Accuracy: 0.5138532834389381
ROC_AUC score: 0.6187926946544635

___________________________________________________________________________________




________________________________________cat_34_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________bin_34_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________37_resp________________________________________
Balanced Accuracy: 0.5830973043786685
ROC_AUC score: 0.6499600977105525

___________________________________________________________________________________




________________________________________cat_37_resp________________________________________
Balanced Accuracy: 0.5210064816192035
ROC_AUC score: 0.5626445206576584

___________________________________________________________________________________




________________________________________bin_37_resp________________________________________
Balanced Accuracy: 0.517216479216485
ROC_AUC score: 0.5665701765426026

___________________________________________________________________________________




________________________________________lit_37_resp________________________________________
Balanced Accuracy: 0.5786410686683147
ROC_AUC score: 0.6238457544171984

___________________________________________________________________________________




________________________________________ola_37_resp________________________________________
Balanced Accuracy: 0.5715256733541256
ROC_AUC score: 0.6935510628414695

___________________________________________________________________________________




________________________________________37_exp________________________________________
Balanced Accuracy: 0.5539152712838864
ROC_AUC score: 0.5994841445594751

___________________________________________________________________________________




________________________________________cat_37_exp________________________________________
Balanced Accuracy: 0.5491100712496624
ROC_AUC score: 0.5844461374697395

___________________________________________________________________________________




________________________________________bin_37_exp________________________________________
Balanced Accuracy: 0.5471165025089209
ROC_AUC score: 0.5855459398429387

___________________________________________________________________________________




________________________________________37_multi________________________________________
Balanced Accuracy: 0.3875424076373301
ROC_AUC score: 0.6095569396838123

___________________________________________________________________________________




________________________________________cat_37_multi________________________________________
Balanced Accuracy: 0.34708807903337985
ROC_AUC score: 0.5440369409509134

___________________________________________________________________________________




________________________________________bin_37_multi________________________________________
Balanced Accuracy: 0.3461819988708128
ROC_AUC score: 0.5478197389143789

___________________________________________________________________________________




________________________________________lit_37_multi________________________________________
Balanced Accuracy: 0.3850145106903992
ROC_AUC score: 0.5951377257494506

___________________________________________________________________________________




________________________________________ola_37_multi________________________________________
Balanced Accuracy: 0.38150595681310495
ROC_AUC score: 0.6409861821257834

___________________________________________________________________________________




________________________________________37_lithium2y________________________________________
Balanced Accuracy: 0.512551877672003
ROC_AUC score: 0.619842045721224

___________________________________________________________________________________




________________________________________cat_37_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________bin_37_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________13_resp________________________________________
Balanced Accuracy: 0.5822266049587533
ROC_AUC score: 0.6433300963375705

___________________________________________________________________________________




________________________________________cat_13_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5031357621766341

___________________________________________________________________________________




________________________________________bin_13_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________lit_13_resp________________________________________
Balanced Accuracy: 0.576376784572234
ROC_AUC score: 0.6225498156535128

___________________________________________________________________________________




________________________________________ola_13_resp________________________________________
Balanced Accuracy: 0.5519614457298325
ROC_AUC score: 0.6735587309311567

___________________________________________________________________________________




________________________________________13_exp________________________________________
Balanced Accuracy: 0.519108292943985
ROC_AUC score: 0.5672112518097616

___________________________________________________________________________________




________________________________________cat_13_exp________________________________________
Balanced Accuracy: 0.5054545542240632
ROC_AUC score: 0.5338737348445577

___________________________________________________________________________________




________________________________________bin_13_exp________________________________________
Balanced Accuracy: 0.5003936178957946
ROC_AUC score: 0.5499176485094946

___________________________________________________________________________________




________________________________________13_multi________________________________________
Balanced Accuracy: 0.3878394785017754
ROC_AUC score: 0.6070247192189846

___________________________________________________________________________________




________________________________________cat_13_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.5003737582196537

___________________________________________________________________________________




________________________________________bin_13_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________lit_13_multi________________________________________
Balanced Accuracy: 0.38402994192011985
ROC_AUC score: 0.5939708088740229

___________________________________________________________________________________




________________________________________ola_13_multi________________________________________
Balanced Accuracy: 0.36855795482750064
ROC_AUC score: 0.6250043455668536

___________________________________________________________________________________




________________________________________13_lithium2y________________________________________
Balanced Accuracy: 0.5084324333976145
ROC_AUC score: 0.6024577644964135

___________________________________________________________________________________




________________________________________cat_13_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________bin_13_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5

___________________________________________________________________________________




________________________________________num_34_resp________________________________________
Balanced Accuracy: 0.5824969107904944
ROC_AUC score: 0.6438724242285558

___________________________________________________________________________________




________________________________________num_34_exp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5455372076910043

___________________________________________________________________________________




________________________________________num_34_multi________________________________________
Balanced Accuracy: 0.38732593956959604
ROC_AUC score: 0.6073454289552751

___________________________________________________________________________________




________________________________________num_34_lithium2y________________________________________
Balanced Accuracy: 0.5045200853000296
ROC_AUC score: 0.6013688754374688

___________________________________________________________________________________




________________________________________num_37_resp________________________________________
Balanced Accuracy: 0.5860582258784225
ROC_AUC score: 0.64826060342559

___________________________________________________________________________________




________________________________________num_37_exp________________________________________
Balanced Accuracy: 0.5007556750599339
ROC_AUC score: 0.5560399061859873

___________________________________________________________________________________




________________________________________num_37_multi________________________________________
Balanced Accuracy: 0.3902299311659596
ROC_AUC score: 0.6105212605735918

___________________________________________________________________________________




________________________________________num_37_lithium2y________________________________________
Balanced Accuracy: 0.5076257308151458
ROC_AUC score: 0.6052576282816533

___________________________________________________________________________________




________________________________________num_13_resp________________________________________
Balanced Accuracy: 0.5824969107904944
ROC_AUC score: 0.6438352392993214

___________________________________________________________________________________




________________________________________num_13_exp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5454701018163882

___________________________________________________________________________________




________________________________________num_13_multi________________________________________
Balanced Accuracy: 0.38732593956959604
ROC_AUC score: 0.6073560975471447

___________________________________________________________________________________




________________________________________num_13_lithium2y________________________________________
Balanced Accuracy: 0.5044206026100176
ROC_AUC score: 0.6013454396114563

___________________________________________________________________________________

display(all_results)

	features	balanced accuracy	accuracy	roc_auc	f1 (response)	f1 (equivocal)	f1 (no response)	f1 (lithium)	f1 (olanzapine)	f1 (lithium > 2y)	f1 (other)	f1 (weighted avg)
0	34_resp	0.580700	0.600151	0.644544	0.416942	NaN	0.695752	NaN	NaN	NaN	NaN	0.566286
1	cat_34_resp	0.520681	0.551113	0.562743	0.162562	NaN	0.693378	NaN	NaN	NaN	NaN	0.446892
2	bin_34_resp	0.517216	0.548849	0.566570	0.131445	NaN	0.695287	NaN	NaN	NaN	NaN	0.433465
3	lit_34_resp	0.576377	0.561359	0.622536	0.527458	NaN	0.590722	NaN	NaN	NaN	NaN	0.555586
4	ola_34_resp	0.562326	0.635294	0.685271	0.297371	NaN	0.753734	NaN	NaN	NaN	NaN	0.570544
5	34_exp	0.556729	0.522952	0.597592	NaN	NaN	NaN	0.564147	0.473156	NaN	NaN	0.510661
6	cat_34_exp	0.552833	0.517692	0.585101	NaN	NaN	NaN	0.562717	0.462331	NaN	NaN	0.503708
7	bin_34_exp	0.547117	0.510360	0.585546	NaN	NaN	NaN	0.560137	0.447879	NaN	NaN	0.494150
8	34_multi	0.385773	0.509882	0.607574	0.394888	0.0	0.626716	NaN	NaN	NaN	NaN	0.442636
9	cat_34_multi	0.347006	0.469238	0.544644	0.158200	0.0	0.623879	NaN	NaN	NaN	NaN	0.347330
10	bin_34_multi	0.346182	0.468441	0.547820	0.150086	0.0	0.623939	NaN	NaN	NaN	NaN	0.344135
11	lit_34_multi	0.384124	0.471301	0.593962	0.491928	0.0	0.528982	NaN	NaN	NaN	NaN	0.427151
12	ola_34_multi	0.374968	0.542003	0.634822	0.290523	0.0	0.680728	NaN	NaN	NaN	NaN	0.447247
13	34_lithium2y	0.513853	0.801084	0.618793	NaN	NaN	NaN	NaN	NaN	0.068657	0.888651	0.725541
14	cat_34_lithium2y	0.500000	0.801084	0.500000	NaN	NaN	NaN	NaN	NaN	0.000000	0.889558	0.712610
15	bin_34_lithium2y	0.500000	0.801084	0.500000	NaN	NaN	NaN	NaN	NaN	0.000000	0.889558	0.712610
16	37_resp	0.583097	0.602603	0.649960	0.419719	NaN	0.697835	NaN	NaN	NaN	NaN	0.568691
17	cat_37_resp	0.521006	0.551867	0.562645	0.154448	NaN	0.695150	NaN	NaN	NaN	NaN	0.444074
18	bin_37_resp	0.517216	0.548849	0.566570	0.131445	NaN	0.695287	NaN	NaN	NaN	NaN	0.433465
19	lit_37_resp	0.578641	0.563247	0.623846	0.527891	NaN	0.593677	NaN	NaN	NaN	NaN	0.557141
20	ola_37_resp	0.571526	0.643765	0.693551	0.316170	NaN	0.759147	NaN	NaN	NaN	NaN	0.581331
21	37_exp	0.553915	0.518489	0.599484	NaN	NaN	NaN	0.564006	0.462360	NaN	NaN	0.504256
22	cat_37_exp	0.549110	0.512432	0.584446	NaN	NaN	NaN	0.561685	0.450709	NaN	NaN	0.496451
23	bin_37_exp	0.547117	0.510360	0.585546	NaN	NaN	NaN	0.560137	0.447879	NaN	NaN	0.494150
24	37_multi	0.387542	0.512113	0.609557	0.398541	0.0	0.628659	NaN	NaN	NaN	NaN	0.444973
25	cat_37_multi	0.347088	0.469398	0.544037	0.157211	0.0	0.624066	NaN	NaN	NaN	NaN	0.347023
26	bin_37_multi	0.346182	0.468441	0.547820	0.150086	0.0	0.623939	NaN	NaN	NaN	NaN	0.344135
27	lit_37_multi	0.385015	0.471827	0.595138	0.490578	0.0	0.530845	NaN	NaN	NaN	NaN	0.427218
28	ola_37_multi	0.381506	0.548869	0.640986	0.313316	0.0	0.685266	NaN	NaN	NaN	NaN	0.457389
29	37_lithium2y	0.512552	0.800446	0.619842	NaN	NaN	NaN	NaN	NaN	0.064275	0.888314	0.724399
30	cat_37_lithium2y	0.500000	0.801084	0.500000	NaN	NaN	NaN	NaN	NaN	0.000000	0.889558	0.712610
31	bin_37_lithium2y	0.500000	0.801084	0.500000	NaN	NaN	NaN	NaN	NaN	0.000000	0.889558	0.712610
32	13_resp	0.582227	0.599585	0.643330	0.439989	NaN	0.688390	NaN	NaN	NaN	NaN	0.573044
33	cat_13_resp	0.500000	0.535647	0.503136	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
34	bin_13_resp	0.500000	0.535647	0.500000	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
35	lit_13_resp	0.576377	0.561359	0.622550	0.527458	NaN	0.590722	NaN	NaN	NaN	NaN	0.555586
36	ola_13_resp	0.551961	0.626353	0.673559	0.272894	NaN	0.748575	NaN	NaN	NaN	NaN	0.557631
37	13_exp	0.519108	0.441026	0.567211	NaN	NaN	NaN	0.586974	0.135568	NaN	NaN	0.321628
38	cat_13_exp	0.505455	0.421103	0.533874	NaN	NaN	NaN	0.583963	0.048717	NaN	NaN	0.269333
39	bin_13_exp	0.500394	0.413931	0.549918	NaN	NaN	NaN	0.582681	0.016056	NaN	NaN	0.249606
40	13_multi	0.387839	0.510679	0.607025	0.417533	0.0	0.622216	NaN	NaN	NaN	NaN	0.449578
41	cat_13_multi	0.333333	0.456009	0.500374	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
42	bin_13_multi	0.333333	0.456009	0.500000	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
43	lit_13_multi	0.384030	0.471301	0.593971	0.492570	0.0	0.528489	NaN	NaN	NaN	NaN	0.427267
44	ola_13_multi	0.368558	0.535137	0.625004	0.265834	0.0	0.678005	NaN	NaN	NaN	NaN	0.437382
45	13_lithium2y	0.508432	0.801084	0.602458	NaN	NaN	NaN	NaN	NaN	0.042945	0.889007	0.720712
46	cat_13_lithium2y	0.500000	0.801084	0.500000	NaN	NaN	NaN	NaN	NaN	0.000000	0.889558	0.712610
47	bin_13_lithium2y	0.500000	0.801084	0.500000	NaN	NaN	NaN	NaN	NaN	0.000000	0.889558	0.712610
48	num_34_resp	0.582497	0.599585	0.643872	0.442928	NaN	0.687472	NaN	NaN	NaN	NaN	0.573918
49	num_34_exp	0.500000	0.412177	0.545537	NaN	NaN	NaN	0.583747	0.000000	NaN	NaN	0.240607
50	num_34_multi	0.387326	0.509882	0.607345	0.418582	0.0	0.620580	NaN	NaN	NaN	NaN	0.449249
51	num_34_lithium2y	0.504520	0.800606	0.601369	NaN	NaN	NaN	NaN	NaN	0.024942	0.888948	0.717083
52	num_37_resp	0.586058	0.603169	0.648261	0.447479	NaN	0.690406	NaN	NaN	NaN	NaN	0.577602
53	num_37_exp	0.500756	0.413134	0.556040	NaN	NaN	NaN	0.584049	0.003788	NaN	NaN	0.242958
54	num_37_multi	0.390230	0.513548	0.610521	0.425006	0.0	0.623219	NaN	NaN	NaN	NaN	0.453004
55	num_37_lithium2y	0.507626	0.801721	0.605258	NaN	NaN	NaN	NaN	NaN	0.037152	0.889481	0.719939
56	num_13_resp	0.582497	0.599585	0.643835	0.442928	NaN	0.687472	NaN	NaN	NaN	NaN	0.573918
57	num_13_exp	0.500000	0.412177	0.545470	NaN	NaN	NaN	0.583747	0.000000	NaN	NaN	0.240607
58	num_13_multi	0.387326	0.509882	0.607356	0.418582	0.0	0.620580	NaN	NaN	NaN	NaN	0.449249
59	num_13_lithium2y	0.504421	0.800446	0.601345	NaN	NaN	NaN	NaN	NaN	0.024922	0.888849	0.717000

SHAP

import shap

def shap_plot(clf, X, sample=50):
    explainer = shap.KernelExplainer(clf.predict, shap.sample(X, sample))
    shap_values = explainer.shap_values(shap.sample(X, sample), l1_reg="num_features("+str(X.shape[1])+")")
    shap.summary_plot(shap_values, shap.sample(X, sample), feature_names=X.columns)

shap_plot(clf_multi_dict['13_lithium2y'], X_dict['13_lithium2y'], 100)

X does not have valid feature names, but LogisticRegressionCV was fitted with feature names

  0%|          | 0/100 [00:00<?, ?it/s]

X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names
X does not have valid feature names, but LogisticRegressionCV was fitted with feature names

UMAP

Inspired by: https://github.com/lmcinnes/umap/issues/58 More precisely: https://github.com/lmcinnes/umap/issues/58#issuecomment-419682509

def umap_embedding(X_dict, n_neighbors=15, weight=0.5):
    import umap.umap_ as umap
    fit1 = umap.UMAP(n_neighbors=n_neighbors, metric='braycurtis', random_state=42).fit(X_dict['num'].values)
    fit2 = umap.UMAP(n_neighbors=n_neighbors, metric='jaccard', random_state=42).fit(X_dict['bin'].values)
    intersection = umap.general_simplicial_set_intersection(fit1.graph_, fit2.graph_, weight=weight)
    intersection = umap.reset_local_connectivity(intersection)
    embedding = umap.simplicial_set_embedding(fit1._raw_data, intersection, fit1.n_components,
                                              fit1._initial_alpha, fit1._a, fit1._b,
                                              fit1.repulsion_strength, fit1.negative_sample_rate,
                                              200, 'random', np.random, fit1.metric,
                                              fit1._metric_kwds, False,
                                              densmap_kwds={}, output_dens=False)
    return embedding

from mpl_toolkits.mplot3d import Axes3D
import seaborn as sns
sns.set(style='white', context='poster', rc={'figure.figsize':(14,10)})

def plot_umap(X_dict, n_neighbors, weight, min_dist=0.1, n_components=2):
    import umap.umap_ as umap
    import matplotlib.pyplot as plt
    print('n_neighbors={:.0f} min_dist={:.1f} weight={:.1f} {:.0f}D'.format(n_neighbors, min_dist, weight, n_components))
    fit1 = umap.UMAP(n_neighbors=n_neighbors, metric='braycurtis', random_state=42, min_dist=min_dist, n_components=n_components).fit(X_dict['num'].values)
    fit2 = umap.UMAP(n_neighbors=n_neighbors, metric='jaccard', random_state=42, min_dist=min_dist, n_components=n_components).fit(X_dict['bin'].values)
    intersection = umap.general_simplicial_set_intersection(fit1.graph_, fit2.graph_, weight=weight)
    intersection = umap.reset_local_connectivity(intersection)
    embedding = umap.simplicial_set_embedding(fit1._raw_data, intersection, fit1.n_components,
                                              fit1._initial_alpha, fit1._a, fit1._b,
                                              fit1.repulsion_strength, fit1.negative_sample_rate,
                                              200, 'random', np.random, fit1.metric,
                                              fit1._metric_kwds, False,
                                              densmap_kwds={}, output_dens=False
                                              )
    plt.clf()
    fig = plt.figure()
    if n_components == 3:
        ax = fig.add_subplot(111, projection='3d')
        ax.scatter(
            embedding[0][:, 0],
            embedding[0][:, 1],
            embedding[0][:, 2],
            c=[sns.color_palette()[x] for x in X.exposure])
    else:
        ax = fig.add_subplot(111)
        ax.scatter(
            embedding[0][:, 0],
            embedding[0][:, 1],
            c=[sns.color_palette()[x] for x in X.exposure])
    #plt.gca().set_aspect('equal', 'datalim')
    plt.title('n_neighbors={:.0f} min_dist={:.1f} weight={:.1f} {:.0f}D'.format(n_neighbors, min_dist, weight, n_components), fontsize=16)
    plt.savefig('umap_images/n' + str(n_neighbors) + '_md{:.1f}_w{:.1f}_'.format(min_dist, weight) + str(n_components) + 'd.png')

def plot_umap_2y(X_dict, n_neighbors, weight, min_dist=0.1, n_components=2):
    import umap.umap_ as umap
    import matplotlib.pyplot as plt
    import matplotlib.patches as mpatches
    print('n_neighbors={:.0f} min_dist={:.1f} weight={:.1f} {:.0f}D'.format(n_neighbors, min_dist, weight, n_components))
    fit1 = umap.UMAP(n_neighbors=n_neighbors, metric='braycurtis', random_state=42, min_dist=min_dist, n_components=n_components).fit(X_dict['num_13_lithium2y'].values)
    fit2 = umap.UMAP(n_neighbors=n_neighbors, metric='jaccard', random_state=42, min_dist=min_dist, n_components=n_components).fit(X_dict['cat_13_lithium2y'].values)
    intersection = umap.general_simplicial_set_intersection(fit1.graph_, fit2.graph_, weight=weight)
    intersection = umap.reset_local_connectivity(intersection)
    embedding = umap.simplicial_set_embedding(fit1._raw_data, intersection, fit1.n_components,
                                              fit1._initial_alpha, fit1._a, fit1._b,
                                              fit1.repulsion_strength, fit1.negative_sample_rate,
                                              200, 'random', np.random, fit1.metric,
                                              fit1._metric_kwds, False,
                                              densmap_kwds={}, output_dens=False
                                              )
    plt.clf()
    fig = plt.figure(figsize=(36,18))

    if n_components == 3:
        ax = fig.add_subplot(111, projection='3d')
        ax.scatter(embedding[0][:, 0],
                   embedding[0][:, 1],
                   embedding[0][:, 2],
                   c=[sns.color_palette()[col] for col in y_dict['13_lithium2y'].astype(int)])
        classes = ['Lithium for more than 2 years', 'Other']
        class_colours = sns.color_palette()
        recs = []
        for i in range(0, len(class_colours)):
            recs.append(mpatches.Rectangle((0, 0), 1, 1, fc=class_colours[i]))
            ax.legend(recs, classes, loc=1)

    else:
        ax = fig.add_subplot(111)
        ax.scatter(
            embedding[0][:, 0],
            embedding[0][:, 1],
            c=[sns.color_palette()[x] for x in y_dict['13_lithium2y'].astype(int)])
    #plt.gca().set_aspect('equal', 'datalim')
    plt.title('n_neighbors={:.0f} min_dist={:.1f} weight={:.1f} {:.0f}D'.format(n_neighbors, min_dist, weight, n_components), fontsize=16)

    plt.savefig('umap_images/n' + str(n_neighbors) + '_md{:.1f}_w{:.1f}_'.format(min_dist, weight) + str(n_components) + 'd_2y.png')
    return embedding

embedding = plot_umap_2y(X_dict, 50, 0, 0.1, 3)

n_neighbors=50 min_dist=0.1 weight=0.0 3D

gradient function is not yet implemented for jaccard distance metric; inverse_transform will be unavailable
Failed to correctly find n_neighbors for some samples.Results may be less than ideal. Try re-running withdifferent parameters.
A few of your vertices were disconnected from the manifold.  This shouldn't cause problems.
Disconnection_distance = 1 has removed 250 edges.
It has only fully disconnected 5 vertices.
Use umap.utils.disconnected_vertices() to identify them.

<Figure size 1008x720 with 0 Axes>

for n_neighbors in range(50, 250, 20):
    for weight in np.linspace(0, 1, 5):
        for n_components in range(2,4):
            for min_dist in np.linspace(.1, .9, 4):
                plot_umap_2y(X_dict, n_neighbors, weight, min_dist=min_dist, n_components=n_components)
print('done')

Silhouette Score

from sklearn.metrics import silhouette_samples, silhouette_score

PCA

from sklearn.decomposition import PCA
from sklearn.pipeline import Pipeline
pca = PCA(n_components=9)

pipe = Pipeline([('pca', pca), ('logistic', clf_multi_dict['all34'])])
pipe.fit(X_dict['all34'], y)
#predictions = pipe.predict(X)

plt.figure(1, figsize=(8, 6))
plt.clf()
plt.plot(pca.explained_variance_, linewidth=2)
plt.axis('tight')
plt.xlabel('n components')
plt.ylabel('explained variance')

pd.set_option('display.float_format', lambda x: '%.3f' % x)
display(pd.DataFrame(pca.components_,columns=X_dict['all34'].columns))

Random Forest

from sklearn.ensemble import RandomForestClassifier

@ignore_warnings(category=ConvergenceWarning) # max_iter default value (=100?) triggers this warning
def run(X1, y):
    X = X1
    if isinstance(X1, pd.Series):
        X = X1.to_frame()
    return RandomForestClassifier(max_depth=2, random_state=0).fit(X, y)

clf2_multi_dict = dict()
for v in X_dict:
    print(v)
    clf2_multi_dict.update({v: run(X_dict[v], y_dict[v])})

34_resp
num_34_resp
cat_34_resp
bin_34_resp
lit_34_resp
ola_34_resp
34_exp
num_34_exp
cat_34_exp
bin_34_exp
34_multi
num_34_multi
cat_34_multi
bin_34_multi
lit_34_multi
ola_34_multi
34_lithium2y
num_34_lithium2y
cat_34_lithium2y
bin_34_lithium2y
37_resp
num_37_resp
cat_37_resp
bin_37_resp
lit_37_resp
ola_37_resp
37_exp
num_37_exp
cat_37_exp
bin_37_exp
37_multi
num_37_multi
cat_37_multi
bin_37_multi
lit_37_multi
ola_37_multi
37_lithium2y
num_37_lithium2y
cat_37_lithium2y
bin_37_lithium2y
13_resp
num_13_resp
cat_13_resp
bin_13_resp
lit_13_resp
ola_13_resp
13_exp
num_13_exp
cat_13_exp
bin_13_exp
13_multi
num_13_multi
cat_13_multi
bin_13_multi
lit_13_multi
ola_13_multi
13_lithium2y
num_13_lithium2y
cat_13_lithium2y
bin_13_lithium2y

Results

all_results2 = evaluate(clf2_multi_dict, x_test, y_test)

________________________________________34_resp________________________________________
Balanced Accuracy: 0.5110331975606687
ROC_AUC score: 0.6441694746055537

___________________________________________________________________________________




________________________________________num_34_resp________________________________________
Balanced Accuracy: 0.5773642464045033
ROC_AUC score: 0.6390544015514696

___________________________________________________________________________________




________________________________________cat_34_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5785016761821947

___________________________________________________________________________________




________________________________________bin_34_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5758893633940115

___________________________________________________________________________________




________________________________________lit_34_resp________________________________________
Balanced Accuracy: 0.6069482286096927
ROC_AUC score: 0.6300764451501537

___________________________________________________________________________________




________________________________________ola_34_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.6840147979384636

___________________________________________________________________________________




________________________________________34_exp________________________________________
Balanced Accuracy: 0.508012755987941
ROC_AUC score: 0.582772370162715

___________________________________________________________________________________




________________________________________num_34_exp________________________________________
Balanced Accuracy: 0.5187031412259908
ROC_AUC score: 0.5527076800995852

___________________________________________________________________________________




________________________________________cat_34_exp________________________________________
Balanced Accuracy: 0.509346904657651
ROC_AUC score: 0.5603262939270862

___________________________________________________________________________________




________________________________________bin_34_exp________________________________________
Balanced Accuracy: 0.5144463220108947
ROC_AUC score: 0.5749915488539155

___________________________________________________________________________________




________________________________________34_multi________________________________________
Balanced Accuracy: 0.34516011062967616
ROC_AUC score: 0.611353800587708

___________________________________________________________________________________




________________________________________num_34_multi________________________________________
Balanced Accuracy: 0.3847279954832512
ROC_AUC score: 0.605524634439378

___________________________________________________________________________________




________________________________________cat_34_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.56156310061291

___________________________________________________________________________________




________________________________________bin_34_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.5625488160989097

___________________________________________________________________________________




________________________________________lit_34_multi________________________________________
Balanced Accuracy: 0.40293891192888515
ROC_AUC score: 0.6005018451992608

___________________________________________________________________________________




________________________________________ola_34_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.6420242532566227

___________________________________________________________________________________




________________________________________34_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.6211246390563939

___________________________________________________________________________________




________________________________________num_34_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.6013454396114563

___________________________________________________________________________________




________________________________________cat_34_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5904663538063608

___________________________________________________________________________________




________________________________________bin_34_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5786242468650198

___________________________________________________________________________________




________________________________________37_resp________________________________________
Balanced Accuracy: 0.5845093305568586
ROC_AUC score: 0.6580121365888261

___________________________________________________________________________________




________________________________________num_37_resp________________________________________
Balanced Accuracy: 0.6025075514010136
ROC_AUC score: 0.6566489513849956

___________________________________________________________________________________




________________________________________cat_37_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5794025526023729

___________________________________________________________________________________




________________________________________bin_37_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5777174317227491

___________________________________________________________________________________




________________________________________lit_37_resp________________________________________
Balanced Accuracy: 0.601644382694887
ROC_AUC score: 0.6358305816688019

___________________________________________________________________________________




________________________________________ola_37_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.6925096035450168

___________________________________________________________________________________




________________________________________37_exp________________________________________
Balanced Accuracy: 0.5125252066441527
ROC_AUC score: 0.581302279670443

___________________________________________________________________________________




________________________________________num_37_exp________________________________________
Balanced Accuracy: 0.519850861387783
ROC_AUC score: 0.5650832616139299

___________________________________________________________________________________




________________________________________cat_37_exp________________________________________
Balanced Accuracy: 0.5111294044521393
ROC_AUC score: 0.5589588020259263

___________________________________________________________________________________




________________________________________bin_37_exp________________________________________
Balanced Accuracy: 0.5129434649782829
ROC_AUC score: 0.5686375662041394

___________________________________________________________________________________




________________________________________37_multi________________________________________
Balanced Accuracy: 0.38740060468928744
ROC_AUC score: 0.6205182267712338

___________________________________________________________________________________




________________________________________num_37_multi________________________________________
Balanced Accuracy: 0.4014441459086413
ROC_AUC score: 0.6173249209650806

___________________________________________________________________________________




________________________________________cat_37_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.5630210871861139

___________________________________________________________________________________




________________________________________bin_37_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.56365325092559

___________________________________________________________________________________




________________________________________lit_37_multi________________________________________
Balanced Accuracy: 0.4018588955331041
ROC_AUC score: 0.6062636060009988

___________________________________________________________________________________




________________________________________ola_37_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.6479986686804441

___________________________________________________________________________________




________________________________________37_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.6199223971246952

___________________________________________________________________________________




________________________________________num_37_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.60939492842348

___________________________________________________________________________________




________________________________________cat_37_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5880130054485904

___________________________________________________________________________________




________________________________________bin_37_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5778896851755487

___________________________________________________________________________________




________________________________________13_resp________________________________________
Balanced Accuracy: 0.5477520280088328
ROC_AUC score: 0.637369066143408

___________________________________________________________________________________




________________________________________num_13_resp________________________________________
Balanced Accuracy: 0.5788939771856157
ROC_AUC score: 0.6401284024210251

___________________________________________________________________________________




________________________________________cat_13_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5358554965046167

___________________________________________________________________________________




________________________________________bin_13_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5100019593597326

___________________________________________________________________________________




________________________________________lit_13_resp________________________________________
Balanced Accuracy: 0.5995958210786525
ROC_AUC score: 0.6204130403838095

___________________________________________________________________________________




________________________________________ola_13_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.674893734285946

___________________________________________________________________________________




________________________________________13_exp________________________________________
Balanced Accuracy: 0.5095903731589922
ROC_AUC score: 0.552450056452817

___________________________________________________________________________________




________________________________________num_13_exp________________________________________
Balanced Accuracy: 0.5182964167140602
ROC_AUC score: 0.5508419270794014

___________________________________________________________________________________




________________________________________cat_13_exp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5219919057732861

___________________________________________________________________________________




________________________________________bin_13_exp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5737315312050705

___________________________________________________________________________________




________________________________________13_multi________________________________________
Balanced Accuracy: 0.36479965529680997
ROC_AUC score: 0.605131810802567

___________________________________________________________________________________




________________________________________num_13_multi________________________________________
Balanced Accuracy: 0.38581131764253085
ROC_AUC score: 0.6047799299462205

___________________________________________________________________________________




________________________________________cat_13_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.5304525184350798

___________________________________________________________________________________




________________________________________bin_13_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.5108637439107558

___________________________________________________________________________________




________________________________________lit_13_multi________________________________________
Balanced Accuracy: 0.3999337699531935
ROC_AUC score: 0.593113733735595

___________________________________________________________________________________




________________________________________ola_13_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.6325575057349074

___________________________________________________________________________________




________________________________________13_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.6051245861265011

___________________________________________________________________________________




________________________________________num_13_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.6015531734978113

___________________________________________________________________________________




________________________________________cat_13_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5407494809044252

___________________________________________________________________________________




________________________________________bin_13_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5400218543063251

___________________________________________________________________________________

display(all_results2)

	features	balanced accuracy	accuracy	roc_auc	f1 (response)	f1 (equivocal)	f1 (no response)	f1 (lithium)	f1 (olanzapine)	f1 (lithium > 2y)	f1 (other)	f1 (weighted avg)
0	34_resp	0.511033	0.545266	0.644169	0.059306	NaN	0.700162	NaN	NaN	NaN	NaN	0.402578
1	num_34_resp	0.577364	0.594115	0.639054	0.439291	NaN	0.681939	NaN	NaN	NaN	NaN	0.569265
2	cat_34_resp	0.500000	0.535647	0.578502	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
3	bin_34_resp	0.500000	0.535647	0.575889	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
4	lit_34_resp	0.606948	0.610132	0.630076	0.644272	NaN	0.568743	NaN	NaN	NaN	NaN	0.610691
5	ola_34_resp	0.500000	0.598588	0.684015	0.000000	NaN	0.748896	NaN	NaN	NaN	NaN	0.448280
6	34_exp	0.508013	0.424450	0.582772	NaN	NaN	NaN	0.584895	0.061834	NaN	NaN	0.277428
7	num_34_exp	0.518703	0.436882	0.552708	NaN	NaN	NaN	0.590377	0.099414	NaN	NaN	0.301778
8	cat_34_exp	0.509347	0.429551	0.560326	NaN	NaN	NaN	0.582039	0.101882	NaN	NaN	0.299792
9	bin_34_exp	0.514446	0.440708	0.574992	NaN	NaN	NaN	0.579307	0.165914	NaN	NaN	0.336305
10	34_multi	0.345160	0.469238	0.611354	0.093006	0.0	0.630719	NaN	NaN	NaN	NaN	0.324555
11	num_34_multi	0.384728	0.505738	0.605525	0.421562	0.0	0.614419	NaN	NaN	NaN	NaN	0.447623
12	cat_34_multi	0.333333	0.456009	0.561563	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
13	bin_34_multi	0.333333	0.456009	0.562549	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
14	lit_34_multi	0.402939	0.512112	0.600502	0.597719	0.0	0.505909	NaN	NaN	NaN	NaN	0.467889
15	ola_34_multi	0.333333	0.510501	0.642024	0.000000	0.0	0.675936	NaN	NaN	NaN	NaN	0.345066
16	34_lithium2y	0.500000	0.801084	0.621125	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
17	num_34_lithium2y	0.500000	0.801084	0.601345	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
18	cat_34_lithium2y	0.500000	0.801084	0.590466	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
19	bin_34_lithium2y	0.500000	0.801084	0.578624	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
20	37_resp	0.584509	0.605998	0.658012	0.400230	NaN	0.706642	NaN	NaN	NaN	NaN	0.564359
21	num_37_resp	0.602508	0.618069	0.656649	0.483023	NaN	0.697174	NaN	NaN	NaN	NaN	0.597732
22	cat_37_resp	0.500000	0.535647	0.579403	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
23	bin_37_resp	0.500000	0.535647	0.577717	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
24	lit_37_resp	0.601644	0.606671	0.635831	0.646293	NaN	0.557052	NaN	NaN	NaN	NaN	0.606615
25	ola_37_resp	0.500000	0.598588	0.692510	0.000000	NaN	0.748896	NaN	NaN	NaN	NaN	0.448280
26	37_exp	0.512525	0.433695	0.581302	NaN	NaN	NaN	0.583226	0.116828	NaN	NaN	0.309067
27	num_37_exp	0.519851	0.443258	0.565083	NaN	NaN	NaN	0.585990	0.150328	NaN	NaN	0.329898
28	cat_37_exp	0.511129	0.431782	0.558959	NaN	NaN	NaN	0.582797	0.109418	NaN	NaN	0.304534
29	bin_37_exp	0.512943	0.436564	0.568638	NaN	NaN	NaN	0.581012	0.140112	NaN	NaN	0.321841
30	37_multi	0.387401	0.513548	0.620518	0.379188	0.0	0.634992	NaN	NaN	NaN	NaN	0.440174
31	num_37_multi	0.401444	0.526458	0.617325	0.458353	0.0	0.630303	NaN	NaN	NaN	NaN	0.469479
32	cat_37_multi	0.333333	0.456009	0.563021	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
33	bin_37_multi	0.333333	0.456009	0.563653	0.000000	0.0	0.626451	NaN	NaN	NaN	NaN	0.285667
34	lit_37_multi	0.401859	0.511058	0.606264	0.596645	0.0	0.504175	NaN	NaN	NaN	NaN	0.466740
35	ola_37_multi	0.333333	0.510501	0.647999	0.000000	0.0	0.675936	NaN	NaN	NaN	NaN	0.345066
36	37_lithium2y	0.500000	0.801084	0.619922	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
37	num_37_lithium2y	0.500000	0.801084	0.609395	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
38	cat_37_lithium2y	0.500000	0.801084	0.588013	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
39	bin_37_lithium2y	0.500000	0.801084	0.577890	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
40	13_resp	0.547752	0.572614	0.637369	0.301910	NaN	0.692036	NaN	NaN	NaN	NaN	0.510880
41	num_13_resp	0.578894	0.595436	0.640128	0.443291	NaN	0.682269	NaN	NaN	NaN	NaN	0.571299
42	cat_13_resp	0.500000	0.535647	0.535855	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
43	bin_13_resp	0.500000	0.535647	0.510002	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
44	lit_13_resp	0.599596	0.606042	0.620413	0.649692	NaN	0.549964	NaN	NaN	NaN	NaN	0.605351
45	ola_13_resp	0.500000	0.598588	0.674894	0.000000	NaN	0.748896	NaN	NaN	NaN	NaN	0.448280
46	13_exp	0.509590	0.424131	0.552450	NaN	NaN	NaN	0.587792	0.044938	NaN	NaN	0.268690
47	num_13_exp	0.518296	0.436404	0.550842	NaN	NaN	NaN	0.590172	0.097959	NaN	NaN	0.300838
48	cat_13_exp	0.500000	0.412177	0.521992	NaN	NaN	NaN	0.583747	0.000000	NaN	NaN	0.240607
49	bin_13_exp	0.500000	0.412177	0.573732	NaN	NaN	NaN	0.583747	0.000000	NaN	NaN	0.240607
50	13_multi	0.364800	0.487727	0.605132	0.285375	0.0	0.624471	NaN	NaN	NaN	NaN	0.398113
51	num_13_multi	0.385811	0.507810	0.604780	0.417036	0.0	0.618374	NaN	NaN	NaN	NaN	0.447629
52	cat_13_multi	0.333333	0.456009	0.530453	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
53	bin_13_multi	0.333333	0.456009	0.510864	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
54	lit_13_multi	0.399934	0.509215	0.593114	0.596050	0.0	0.499676	NaN	NaN	NaN	NaN	0.464782
55	ola_13_multi	0.333333	0.510501	0.632558	0.000000	0.0	0.675936	NaN	NaN	NaN	NaN	0.345066
56	13_lithium2y	0.500000	0.801084	0.605125	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
57	num_13_lithium2y	0.500000	0.801084	0.601553	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
58	cat_13_lithium2y	0.500000	0.801084	0.540749	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610
59	bin_13_lithium2y	0.500000	0.801084	0.540022	NaN	NaN	NaN	NaN	NaN	0.0	0.889558	0.712610

Feature Importance

for v in clf2_multi_dict.keys():
    importances = clf2_multi_dict[v].feature_importances_
    indices = np.argsort(importances)
    features = X_dict[v].columns
    plt.figure(1, figsize=(8, 10))
    plt.clf()
    plt.title('Feature Importances ' + v)
    plt.barh(range(len(indices)), importances[indices], color='b', align='center')
    plt.yticks(range(len(indices)), [features[i] for i in indices])
    plt.xlabel('Relative Importance')
    plt.show()

SHAP

import shap
for v in clf2_multi_dict.keys():
    print(40*'_' + v + 40*'_')
    shap_values = shap.TreeExplainer(clf2_multi_dict[v]).shap_values(X_dict[v])
    shap.summary_plot(shap_values, X_dict[v], plot_type="bar")
    print(83*'_')
    print('\n\n\n')

________________________________________34_resp________________________________________

___________________________________________________________________________________




________________________________________num_34_resp________________________________________

___________________________________________________________________________________




________________________________________cat_34_resp________________________________________

___________________________________________________________________________________




________________________________________bin_34_resp________________________________________

___________________________________________________________________________________




________________________________________lit_34_resp________________________________________

___________________________________________________________________________________




________________________________________ola_34_resp________________________________________

___________________________________________________________________________________




________________________________________34_exp________________________________________

___________________________________________________________________________________




________________________________________num_34_exp________________________________________

___________________________________________________________________________________




________________________________________cat_34_exp________________________________________

___________________________________________________________________________________




________________________________________bin_34_exp________________________________________

___________________________________________________________________________________




________________________________________34_multi________________________________________

___________________________________________________________________________________




________________________________________num_34_multi________________________________________

___________________________________________________________________________________




________________________________________cat_34_multi________________________________________

___________________________________________________________________________________




________________________________________bin_34_multi________________________________________

___________________________________________________________________________________




________________________________________lit_34_multi________________________________________

___________________________________________________________________________________




________________________________________ola_34_multi________________________________________

___________________________________________________________________________________




________________________________________34_lithium2y________________________________________

___________________________________________________________________________________




________________________________________num_34_lithium2y________________________________________

___________________________________________________________________________________




________________________________________cat_34_lithium2y________________________________________

___________________________________________________________________________________




________________________________________bin_34_lithium2y________________________________________

___________________________________________________________________________________




________________________________________37_resp________________________________________

___________________________________________________________________________________




________________________________________num_37_resp________________________________________

___________________________________________________________________________________




________________________________________cat_37_resp________________________________________

___________________________________________________________________________________




________________________________________bin_37_resp________________________________________

___________________________________________________________________________________




________________________________________lit_37_resp________________________________________

___________________________________________________________________________________




________________________________________ola_37_resp________________________________________

___________________________________________________________________________________




________________________________________37_exp________________________________________

___________________________________________________________________________________




________________________________________num_37_exp________________________________________

___________________________________________________________________________________




________________________________________cat_37_exp________________________________________

___________________________________________________________________________________




________________________________________bin_37_exp________________________________________

___________________________________________________________________________________




________________________________________37_multi________________________________________

___________________________________________________________________________________




________________________________________num_37_multi________________________________________

___________________________________________________________________________________




________________________________________cat_37_multi________________________________________

___________________________________________________________________________________




________________________________________bin_37_multi________________________________________

___________________________________________________________________________________




________________________________________lit_37_multi________________________________________

___________________________________________________________________________________




________________________________________ola_37_multi________________________________________

___________________________________________________________________________________




________________________________________37_lithium2y________________________________________

___________________________________________________________________________________




________________________________________num_37_lithium2y________________________________________

___________________________________________________________________________________




________________________________________cat_37_lithium2y________________________________________

___________________________________________________________________________________




________________________________________bin_37_lithium2y________________________________________

___________________________________________________________________________________




________________________________________13_resp________________________________________

___________________________________________________________________________________




________________________________________num_13_resp________________________________________

___________________________________________________________________________________




________________________________________cat_13_resp________________________________________

___________________________________________________________________________________




________________________________________bin_13_resp________________________________________

___________________________________________________________________________________




________________________________________lit_13_resp________________________________________

___________________________________________________________________________________




________________________________________ola_13_resp________________________________________

___________________________________________________________________________________




________________________________________13_exp________________________________________

___________________________________________________________________________________




________________________________________num_13_exp________________________________________

___________________________________________________________________________________




________________________________________cat_13_exp________________________________________

___________________________________________________________________________________




________________________________________bin_13_exp________________________________________

___________________________________________________________________________________




________________________________________13_multi________________________________________

___________________________________________________________________________________




________________________________________num_13_multi________________________________________

___________________________________________________________________________________




________________________________________cat_13_multi________________________________________

___________________________________________________________________________________




________________________________________bin_13_multi________________________________________

___________________________________________________________________________________




________________________________________lit_13_multi________________________________________

___________________________________________________________________________________




________________________________________ola_13_multi________________________________________

___________________________________________________________________________________




________________________________________13_lithium2y________________________________________

___________________________________________________________________________________




________________________________________num_13_lithium2y________________________________________

___________________________________________________________________________________




________________________________________cat_13_lithium2y________________________________________

___________________________________________________________________________________




________________________________________bin_13_lithium2y________________________________________

___________________________________________________________________________________

Naive Bayes

from sklearn.naive_bayes import GaussianNB, BernoulliNB
from mixed_naive_bayes import MixedNB
clf3_dict = dict()


for v in [feats for feats in X_dict if ('_13' not in feats) and ('_3' not in feats)]:
    print(v)
    gnb = GaussianNB()
    v_num = 'num_' + v
    clf3_dict[v_num] = gnb.fit(X_dict[v_num], y_dict[v_num])

    bnb = BernoulliNB()
    v_bin = 'bin_' + v
    clf3_dict[v_bin] = bnb.fit(X_dict[v_bin], y_dict[v_bin])

    bin_features = [X_dict[v].columns.get_loc(c) for c in list(X_dict['bin_' + v].columns)]
    mnb = MixedNB(categorical_features=bin_features)
    clf3_dict[v] = mnb.fit(np.array(X_dict[v]), np.array(y_dict[v]))

34_resp
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
34_exp
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
34_multi
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
34_lithium2y
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
37_resp
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
37_exp
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
37_multi
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
37_lithium2y
[2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2]
13_resp
[2 2 2 2 2 2 2 2]
13_exp
[2 2 2 2 2 2 2 2]
13_multi
[2 2 2 2 2 2 2 2]
13_lithium2y
[2 2 2 2 2 2 2 2]

all_results3 = evaluate(clf3_dict, x_test, y_test)

________________________________________num_34_resp________________________________________
Balanced Accuracy: 0.5857668962597682
ROC_AUC score: 0.6380105061726984

___________________________________________________________________________________




________________________________________bin_34_resp________________________________________
Balanced Accuracy: 0.5367480063157173
ROC_AUC score: 0.5760540497248314

___________________________________________________________________________________




________________________________________34_resp________________________________________
Balanced Accuracy: 0.5952593505795128
ROC_AUC score: 0.50115180318303

___________________________________________________________________________________




________________________________________num_34_exp________________________________________
Balanced Accuracy: 0.5224358006485783
ROC_AUC score: 0.540462221070238

___________________________________________________________________________________




________________________________________bin_34_exp________________________________________
Balanced Accuracy: 0.5493286896068099
ROC_AUC score: 0.5724976219355683

___________________________________________________________________________________




________________________________________34_exp________________________________________
Balanced Accuracy: 0.5580377738968214
ROC_AUC score: 0.49567450211635156

___________________________________________________________________________________




________________________________________num_34_multi________________________________________
Balanced Accuracy: 0.39002599299933854
ROC_AUC score: 0.6033948361067034

___________________________________________________________________________________




________________________________________bin_34_multi________________________________________
Balanced Accuracy: 0.3575640114310101
ROC_AUC score: 0.5615403377461362

___________________________________________________________________________________




________________________________________34_multi________________________________________
Balanced Accuracy: 0.3965460299008324
ROC_AUC score: None

___________________________________________________________________________________




________________________________________num_34_lithium2y________________________________________
Balanced Accuracy: 0.5325429561153795
ROC_AUC score: 0.5977664541818442

___________________________________________________________________________________




________________________________________bin_34_lithium2y________________________________________
Balanced Accuracy: 0.5004006410256411
ROC_AUC score: 0.584060401935576

___________________________________________________________________________________




________________________________________34_lithium2y________________________________________
Balanced Accuracy: 0.5462642336771864
ROC_AUC score: 0.5140489008438173

___________________________________________________________________________________




________________________________________num_37_resp________________________________________
Balanced Accuracy: 0.5951753984508186
ROC_AUC score: 0.6473412775597532

___________________________________________________________________________________




________________________________________bin_37_resp________________________________________
Balanced Accuracy: 0.5367480063157173
ROC_AUC score: 0.5760540497248314

___________________________________________________________________________________




________________________________________37_resp________________________________________
Balanced Accuracy: 0.6166868800128145
ROC_AUC score: 0.5057420681685564

___________________________________________________________________________________




________________________________________num_37_exp________________________________________
Balanced Accuracy: 0.5224040302110647
ROC_AUC score: 0.5437385605454366

___________________________________________________________________________________




________________________________________bin_37_exp________________________________________
Balanced Accuracy: 0.5493286896068099
ROC_AUC score: 0.5724976219355683

___________________________________________________________________________________




________________________________________37_exp________________________________________
Balanced Accuracy: 0.5421201555849703
ROC_AUC score: 0.5031892066911268

___________________________________________________________________________________




________________________________________num_37_multi________________________________________
Balanced Accuracy: 0.39620903166493027
ROC_AUC score: 0.610487813737447

___________________________________________________________________________________




________________________________________bin_37_multi________________________________________
Balanced Accuracy: 0.3575640114310101
ROC_AUC score: 0.5615403377461362

___________________________________________________________________________________




________________________________________37_multi________________________________________
Balanced Accuracy: 0.4107821949731159
ROC_AUC score: None

___________________________________________________________________________________




________________________________________num_37_lithium2y________________________________________
Balanced Accuracy: 0.5497151989143633
ROC_AUC score: 0.6024776929198934

___________________________________________________________________________________




________________________________________bin_37_lithium2y________________________________________
Balanced Accuracy: 0.5004006410256411
ROC_AUC score: 0.584060401935576

___________________________________________________________________________________




________________________________________37_lithium2y________________________________________
Balanced Accuracy: 0.5824579175467058
ROC_AUC score: 0.5084447093064781

___________________________________________________________________________________




________________________________________num_13_resp________________________________________
Balanced Accuracy: 0.5857668962597682
ROC_AUC score: 0.6379753235088843

___________________________________________________________________________________




________________________________________bin_13_resp________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5130504084621457

___________________________________________________________________________________




________________________________________13_resp________________________________________
Balanced Accuracy: 0.5832612041052162
ROC_AUC score: 0.4957072144483473

___________________________________________________________________________________




________________________________________num_13_exp________________________________________
Balanced Accuracy: 0.5224358006485783
ROC_AUC score: 0.5404592851882236

___________________________________________________________________________________




________________________________________bin_13_exp________________________________________
Balanced Accuracy: 0.5025319885316061
ROC_AUC score: 0.5606965296196942

___________________________________________________________________________________




________________________________________13_exp________________________________________
Balanced Accuracy: 0.539042931822109
ROC_AUC score: 0.47635325287338964

___________________________________________________________________________________




________________________________________num_13_multi________________________________________
Balanced Accuracy: 0.39002599299933854
ROC_AUC score: 0.6033941180690108

___________________________________________________________________________________




________________________________________bin_13_multi________________________________________
Balanced Accuracy: 0.3333333333333333
ROC_AUC score: 0.5108453613008589

___________________________________________________________________________________




________________________________________13_multi________________________________________
Balanced Accuracy: 0.38824810288881545
ROC_AUC score: None

___________________________________________________________________________________




________________________________________num_13_lithium2y________________________________________
Balanced Accuracy: 0.5325429561153795
ROC_AUC score: 0.5977613525054333

___________________________________________________________________________________




________________________________________bin_13_lithium2y________________________________________
Balanced Accuracy: 0.5
ROC_AUC score: 0.5417565837134082

___________________________________________________________________________________




________________________________________13_lithium2y________________________________________
Balanced Accuracy: 0.5359470497005316
ROC_AUC score: 0.5166446975726223

___________________________________________________________________________________

roc_auc_score returns an error for 34_multi. Looks like predict_proba() for MixedNB returns probas that don't sum up to 1. Strange.

v = '34_multi'
v2 = v
y_score = clf3_dict[v].predict_proba(x_test[v2])
print(y_score[0])
roc_auc_score(y_test[v2], y_score, average='weighted', multi_class='ovr')
#result = results(clf3_dict[v], x_test[v2], y_test[v2], v)

[4.16702351e-06 1.29765617e-06 3.37361391e-06]

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Input In [91], in <cell line: 5>()
      3 y_score = clf3_dict[v].predict_proba(x_test[v2])
      4 print(y_score[0])
----> 5 roc_auc_score(y_test[v2], y_score, average='weighted', multi_class='ovr')

File ~/GoogleDrive/sics/projects/ucl/lithium/venv/lib/python3.9/site-packages/sklearn/metrics/_ranking.py:561, in roc_auc_score(y_true, y_score, average, sample_weight, max_fpr, multi_class, labels)
    559     if multi_class == "raise":
    560         raise ValueError("multi_class must be in ('ovo', 'ovr')")
--> 561     return _multiclass_roc_auc_score(
    562         y_true, y_score, labels, multi_class, average, sample_weight
    563     )
    564 elif y_type == "binary":
    565     labels = np.unique(y_true)

File ~/GoogleDrive/sics/projects/ucl/lithium/venv/lib/python3.9/site-packages/sklearn/metrics/_ranking.py:628, in _multiclass_roc_auc_score(y_true, y_score, labels, multi_class, average, sample_weight)
    626 # validation of the input y_score
    627 if not np.allclose(1, y_score.sum(axis=1)):
--> 628     raise ValueError(
    629         "Target scores need to be probabilities for multiclass "
    630         "roc_auc, i.e. they should sum up to 1.0 over classes"
    631     )
    633 # validation for multiclass parameter specifications
    634 average_options = ("macro", "weighted")

ValueError: Target scores need to be probabilities for multiclass roc_auc, i.e. they should sum up to 1.0 over classes

display(all_results3)

	features	balanced accuracy	accuracy	roc_auc	f1 (response)	f1 (equivocal)	f1 (no response)	f1 (lithium)	f1 (olanzapine)	f1 (lithium > 2y)	f1 (other)	f1 (weighted avg)
0	num_34_resp	0.585767	0.599208	0.638011	0.479294	NaN	0.674230	NaN	NaN	NaN	NaN	0.583711
1	bin_34_resp	0.536748	0.564127	0.576054	0.245511	NaN	0.693542	NaN	NaN	NaN	NaN	0.485497
2	34_resp	0.595259	0.608827	0.501152	0.490167	NaN	0.682681	NaN	NaN	NaN	NaN	0.593286
3	num_34_exp	0.522436	0.472107	0.540462	NaN	NaN	NaN	0.558164	0.344418	NaN	NaN	0.432519
4	bin_34_exp	0.549329	0.529806	0.572498	NaN	NaN	NaN	0.536601	0.522808	NaN	NaN	0.528493
5	34_exp	0.558038	0.546701	0.495675	NaN	NaN	NaN	0.531003	0.561382	NaN	NaN	0.548860
6	num_34_multi	0.390026	0.510201	0.603395	0.451128	0.0	0.610942	NaN	NaN	NaN	NaN	0.457781
7	bin_34_multi	0.357564	0.480236	0.56154	0.236620	0.0	0.625000	NaN	NaN	NaN	NaN	0.378990
8	34_multi	0.396546	0.518648	None	0.462348	0.0	0.618352	NaN	NaN	NaN	NaN	0.465616
9	num_34_lithium2y	0.532543	0.762512	0.597766	NaN	NaN	NaN	NaN	NaN	0.201501	0.860513	0.729425
10	bin_34_lithium2y	0.500401	0.801243	0.58406	NaN	NaN	NaN	NaN	NaN	0.001601	0.889636	0.712992
11	34_lithium2y	0.546264	0.768091	0.514049	NaN	NaN	NaN	NaN	NaN	0.233807	0.863367	0.738138
12	num_37_resp	0.595175	0.607318	0.647341	0.501198	NaN	0.676205	NaN	NaN	NaN	NaN	0.594940
13	bin_37_resp	0.536748	0.564127	0.576054	0.245511	NaN	0.693542	NaN	NaN	NaN	NaN	0.485497
14	37_resp	0.616687	0.619012	0.505742	0.587418	NaN	0.646111	NaN	NaN	NaN	NaN	0.618857
15	num_37_exp	0.522404	0.493465	0.543739	NaN	NaN	NaN	0.527926	0.453576	NaN	NaN	0.484222
16	bin_37_exp	0.549329	0.529806	0.572498	NaN	NaN	NaN	0.536601	0.522808	NaN	NaN	0.528493
17	37_exp	0.542120	0.516576	0.503189	NaN	NaN	NaN	0.539687	0.491022	NaN	NaN	0.511081
18	num_37_multi	0.396209	0.517055	0.610488	0.470327	0.0	0.613356	NaN	NaN	NaN	NaN	0.466507
19	bin_37_multi	0.357564	0.480236	0.56154	0.236620	0.0	0.625000	NaN	NaN	NaN	NaN	0.378990
20	37_multi	0.410782	0.527574	None	0.545080	0.0	0.590033	NaN	NaN	NaN	NaN	0.485564
21	num_37_lithium2y	0.549715	0.743704	0.602478	NaN	NaN	NaN	NaN	NaN	0.261029	0.844967	0.728813
22	bin_37_lithium2y	0.500401	0.801243	0.58406	NaN	NaN	NaN	NaN	NaN	0.001601	0.889636	0.712992
23	37_lithium2y	0.582458	0.687600	0.508445	NaN	NaN	NaN	NaN	NaN	0.341840	0.795193	0.705014
24	num_13_resp	0.585767	0.599208	0.637975	0.479294	NaN	0.674230	NaN	NaN	NaN	NaN	0.583711
25	bin_13_resp	0.500000	0.535647	0.51305	0.000000	NaN	0.697617	NaN	NaN	NaN	NaN	0.373677
26	13_resp	0.583261	0.596379	0.495707	0.478811	NaN	0.670668	NaN	NaN	NaN	NaN	0.581579
27	num_13_exp	0.522436	0.472107	0.540459	NaN	NaN	NaN	0.558164	0.344418	NaN	NaN	0.432519
28	bin_13_exp	0.502532	0.418075	0.560697	NaN	NaN	NaN	0.582122	0.041984	NaN	NaN	0.264616
29	13_exp	0.539043	0.493465	0.476353	NaN	NaN	NaN	0.565134	0.393511	NaN	NaN	0.464250
30	num_13_multi	0.390026	0.510201	0.603394	0.451128	0.0	0.610942	NaN	NaN	NaN	NaN	0.457781
31	bin_13_multi	0.333333	0.456009	0.510845	0.000000	0.0	0.626382	NaN	NaN	NaN	NaN	0.285636
32	13_multi	0.388248	0.507651	None	0.450124	0.0	0.607827	NaN	NaN	NaN	NaN	0.455962
33	num_13_lithium2y	0.532543	0.762512	0.597761	NaN	NaN	NaN	NaN	NaN	0.201501	0.860513	0.729425
34	bin_13_lithium2y	0.500000	0.801084	0.541757	NaN	NaN	NaN	NaN	NaN	0.000000	0.889558	0.712610
35	13_lithium2y	0.535947	0.764106	0.516645	NaN	NaN	NaN	NaN	NaN	0.209402	0.861371	0.731684

Hyperparameter Tuning

LightGBM

import lightgbm as lgb
from sklearn.model_selection import cross_val_score
from hyperopt import fmin, tpe, hp, STATUS_OK, Trials, space_eval
from hyperopt.pyll import scope
import time

param_hyperopt= {
    'learning_rate': hp.loguniform('learning_rate', np.log(0.01), np.log(1)),
    'max_depth': scope.int(hp.quniform('max_depth', 5, 15, 1)),
    'n_estimators': scope.int(hp.quniform('n_estimators', 5, 35, 1)),
    'num_leaves': scope.int(hp.quniform('num_leaves', 5, 50, 1)),
    'boosting_type': hp.choice('boosting_type', ['gbdt', 'dart']),
    'colsample_bytree': hp.uniform('colsample_by_tree', 0.6, 1.0),
    'reg_lambda': hp.uniform('reg_lambda', 0.0, 1.0),
}

def hyperopt_lgbm(param_space, X_train, y_train, X_test, y_test, num_eval, metric):
    
    start = time.time()
    
    def objective_function(params):
        clf = lgb.LGBMClassifier(**params)
        score = cross_val_score(clf, X_train, y_train, cv=5, scoring=metric).mean()
        return {'loss': -score, 'status': STATUS_OK}

    trials = Trials()
    best_param = fmin(objective_function, 
                      param_space, 
                      algo=tpe.suggest, 
                      max_evals=num_eval, 
                      trials=trials)
    loss = [x['result']['loss'] for x in trials.trials]
    best_param_dict = space_eval(param_space, best_param)
    clf_best = lgb.LGBMClassifier(**best_param_dict)
    clf_best.fit(X_train, y_train)
    
    print("")
    print("##### Results")
    print("Score best parameters: ", min(loss)*-1)
    print("Best parameters: ", best_param_dict)
    print("Test Score: ", clf_best.score(X_test, y_test))
    print("Time elapsed: ", time.time() - start)
    print("Parameter combinations evaluated: ", num_eval)

    
    return clf_best, best_param_dict

results_hyperopt_lgbm = dict()
best_params_dict_lgbm = dict()
for target in ['lithium2y']:
    for feature_number in ['13']:
        feature_set = feature_number + '_' + target
        print('\n' + 10*'_' + feature_set + 10*'_')
        for metric in ['balanced_accuracy']:
            print('\n\t----> metric:',  metric)
            results_hyperopt_lgbm[feature_set + '_' + metric], best_params_dict_lgbm[feature_set + '_' + metric] = hyperopt_lgbm(param_hyperopt,
                                                                                                                     x_train[feature_set],
                                                                                                                     y_train[feature_set],
                                                                                                                     x_test[feature_set],
                                                                                                                     y_test[feature_set],
                                                                                                                     1000, metric)

__________13_lithium2y__________

	----> metric: balanced_accuracy
100%|██████████| 1000/1000 [22:38<00:00,  1.36s/trial, best loss: -0.5348214119885094]

##### Results
Score best parameters:  0.5348214119885094
Best parameters:  {'boosting_type': 'gbdt', 'colsample_bytree': 0.600831897129334, 'learning_rate': 0.9933016655780533, 'max_depth': 12, 'n_estimators': 34, 'num_leaves': 50, 'reg_lambda': 0.6221105546939463}
Test Score:  0.7496015301243226
Time elapsed:  1359.1855659484863
Parameter combinations evaluated:  1000

all_results_hyper_lgbm = evaluate(results_hyperopt_lgbm, x_test, y_test)

________________________________________13_lithium2y_balanced_accuracy________________________________________
Balanced Accuracy: 0.5271952832450744
ROC_AUC score: 0.5663205179221893

___________________________________________________________________________________

display(all_results_hyper_lgbm)

	features	balanced accuracy	accuracy	roc_auc	f1 (response)	f1 (equivocal)	f1 (no response)	f1 (lithium)	f1 (olanzapine)	f1 (lithium > 2y)	f1 (other)	f1 (weighted avg)
0	13_lithium2y_balanced_accuracy	0.527195	0.749602	0.566321	NaN	NaN	NaN	NaN	NaN	0.200509	0.851554	0.722051

Random Forest

from sklearn.model_selection import cross_val_score
from hyperopt import fmin, tpe, hp, STATUS_OK, Trials, space_eval
from hyperopt.pyll import scope
import time

param_hyperopt= {
    'max_depth': scope.int(hp.quniform('max_depth', 5, 15, 1)),
    'n_estimators': scope.int(hp.quniform('n_estimators', 5, 35, 1)),
    'min_samples_split':hp.uniform('min_samples_split',0,1),
    'min_samples_leaf':hp.randint('min_samples_leaf',1,10),
    'criterion':hp.choice('criterion', ['gini','entropy']),
    'max_features':hp.choice('max_features',['sqrt', 'log2'])
}


def hyperopt_rf(param_space, X_train, y_train, X_test, y_test, num_eval, metric):
    start = time.time()

    def objective_function(params):
        clf = RandomForestClassifier(**params, random_state=42)

        clf.fit(X_train, y_train)
        score = cross_val_score(clf, X_train, y_train, cv=5, scoring=metric).mean()
        return {'loss': -score, 'status': STATUS_OK}



    trials = Trials()
    best_param = fmin(objective_function,
                  param_space,
                  algo=tpe.suggest,
                  max_evals=num_eval,
                  trials=trials)

    best_param_dict = space_eval(param_space, best_param)
    loss = [x['result']['loss'] for x in trials.trials]
    clf_best = RandomForestClassifier(**best_param_dict)
    clf_best.fit(X_train, y_train)

    print("##### Results")
    print("Score best parameters: ", min(loss)*-1)
    print("Best parameters: ", best_param)
    print("Test Score: ", clf_best.score(X_test, y_test))
    print("Time elapsed: ", time.time() - start)
    print("Parameter combinations evaluated: ", num_eval)
    #results(clf_best, X_test, y_test)

    return clf_best, best_param_dict

results_hyperopt = dict()
best_params_dict = dict()
for target in ['lithium2y']:
    for feature_number in ['13']:
        feature_set = feature_number + '_' + target
        print('\n' + 10*'_' + feature_set + 10*'_')
        for metric in ['balanced_accuracy']:
            print('\n\t----> metric:',  metric)
            results_hyperopt[feature_set + '_' + metric], best_params_dict[feature_set + '_' + metric] = hyperopt_rf(param_hyperopt,
                                                                                                                     x_train[feature_set],
                                                                                                                     y_train[feature_set],
                                                                                                                     x_test[feature_set],
                                                                                                                     y_test[feature_set],
                                                                                                                     1000, metric)

__________13_lithium2y__________

	----> metric: balanced_accuracy
100%|██████████| 1000/1000 [19:17<00:00,  1.16s/trial, best loss: -0.5106134589427425]
##### Results
Score best parameters:  0.5106134589427425
Best parameters:  {'criterion': 0, 'max_depth': 14.0, 'max_features': 0, 'min_samples_leaf': 2, 'min_samples_split': 0.00035845671793269197, 'n_estimators': 16.0}
Test Score:  0.7975773031558814
Time elapsed:  1158.2907378673553
Parameter combinations evaluated:  1000

all_results_hyper_rf = evaluate(results_hyperopt, x_test, y_test)

________________________________________13_lithium2y_balanced_accuracy________________________________________
Balanced Accuracy: 0.5029310725254319
ROC_AUC score: 0.589657897522626

___________________________________________________________________________________

display(all_results_hyper_rf)

	features	balanced accuracy	accuracy	roc_auc	f1 (response)	f1 (equivocal)	f1 (no response)	f1 (lithium)	f1 (olanzapine)	f1 (lithium > 2y)	f1 (other)	f1 (weighted avg)
0	13_lithium2y_balanced_accuracy	0.502931	0.797577	0.589658	NaN	NaN	NaN	NaN	NaN	0.026074	0.887051	0.715789

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.pipeline import Pipeline
from sklearn.svm import LinearSVC

from umap import UMAP

# Classification with a linear SVM
svc = LinearSVC(dual=False, random_state=123)
params_grid = {"C": [10 ** k for k in range(-3, 4)]}
clf = GridSearchCV(svc, params_grid, scoring='balanced_accuracy')
feature_set = '13_lithium2y'
clf.fit(x_train[feature_set], y_train[feature_set])
print(
    "Balanced accuracy on the test set with raw data: {:.3f}".format(clf.score(x_test[feature_set], y_test[feature_set]))
)

# Transformation with UMAP followed by classification with a linear SVM
umap = UMAP(random_state=456)
pipeline = Pipeline([("umap", umap), ("svc", svc)])
params_grid_pipeline = {
    "umap__n_neighbors": [5, 20],
    "umap__n_components": [15, 25, 50],
    "svc__C": [10 ** k for k in range(-3, 4)],
}


clf_pipeline = GridSearchCV(pipeline, params_grid_pipeline, scoring='balanced_accuracy')
clf_pipeline.fit(x_train[feature_set], y_train[feature_set])
print(
    "Balanced accuracy on the test set with UMAP transformation: {:.3f}".format(
        clf_pipeline.score(x_test[feature_set], y_test[feature_set])
    )
)

Balanced accuracy on the test set with raw data: 0.504

OMP: Info #270: omp_set_nested routine deprecated, please use omp_set_max_active_levels instead.

Balanced accuracy on the test set with UMAP transformation: 0.513

print("Balanced accuracy on the test set with raw data: {:.6f}".format(clf.score(x_test[feature_set], y_test[feature_set])),
      "\nBalanced accuracy on the test set with UMAP transformation: {:.6f}".format(clf_pipeline.score(x_test[feature_set], y_test[feature_set])))

Balanced accuracy on the test set with raw data: 0.504214 
Balanced accuracy on the test set with UMAP transformation: 0.513193